CN105791828A - Binary Arithmetic Coder and Its Encoding Method - Google Patents

Abstract

The invention provides a coding interval updating and normalizing module, a coding lower limit updating and normalizing module, a code stream extracting module, a binary arithmetic coder and a coding method thereof. The binary arithmetic encoder comprises a coding interval updating and normalizing module, a coding lower limit updating and normalizing module and a code stream extracting module, wherein the coding interval updating and normalizing module, the coding lower limit updating and normalizing module and the code stream extracting module form a parallel three-level pipeline structure, and the code stream extracting module can process one or more coding symbols in one clock period, so that complete parallel is realized, and the problem of poor parallel in the prior art is solved; in addition, sharing of different coding modes is realized through the selector, the reusability of the circuit is improved, and the hardware cost is reduced; in addition, the code stream extraction method is simple in code stream extraction operation and does not need extra FIFO resources.

Description

Binary Arithmetic Coder and Its Encoding Method

technical field

The present invention relates to the technical field of video coding, in particular to a coding interval update and normalization module, a coding lower limit update and normalization module, and a code stream extraction module, and also relates to a coding interval update and normalization module 1. A binary arithmetic coder and a coding method of a coding lower limit update and normalization module and a code stream extraction module.

Background technique

Binary arithmetic coder (Context-based Adaptive Binary Arithmetic Coder, referred to as CABAC) is an encoder designed according to the principle of binary arithmetic coding. The binary arithmetic coding technology in CABAC is based on the principle of recursive division of intervals. The coding process of CABAC includes: binarization, context modeling and arithmetic coding. Among them, binarization is to map syntax elements into a series of binary bits through simple variable-length mapping. Arithmetic coding is to encode each bit to obtain the output stream of H.264/AVC. Arithmetic coding includes interval subdivision and interval normalization.

Currently, there are three modes of binary arithmetic coding in the H.264/AVC standard, which are a regular coding mode, an equal-probability coding mode (bypass coding mode) and a cut-off coding mode.

Taking the conventional encoding mode as an example, the encoding range (range) is divided into two sub-intervals, namely the interval of low probability characters (LPS) and the interval of high probability characters (MPS). Wherein, the range length of characters with low probability is represented by rLPS, and the range length of characters with high probability is represented by rMPS, rMPS=range-rLPS. In order to reduce the amount of calculation during interval division, the rLPS is obtained by looking up a table in the binary arithmetic coder. The new encoding interval (range) is determined according to whether the current encoding symbol (bin) is a low probability character (LPS) or a high probability character (MPS). In order to keep the coding range within a certain range, the coding range needs to be normalized. In the normal coding mode, the value range of the coding interval (range) is between 2 ⁸ and 2 ⁹ . If the new coding range (range) exceeds the value range, the coding range (range) needs to be normalized to keep it between 2 ⁸ and 2 ⁹ . The normalization process also includes adjusting the encoding lower limit (low) to match the normalized encoding range.

For the equal-probability coding mode, the coding interval (range) is fixed. When the encoding symbol (bin) is "0", the encoding lower limit (low) is 2*low. When the coding symbol (bin) is "1", the coding lower limit (low) is 2*low+range.

For the cut-off encoding mode, the interval length (rLPS) of the low probability character is fixed at 2, and other operations are the same as the regular encoding mode.

It can be seen that the three encoding modes are different from each other. If the traditional binary arithmetic encoder wants to process the three encoding modes at the same time, it is necessary to set up a special logic control circuit and increase additional hardware costs. Moreover, the binary arithmetic coding is a serial coding, and the parallelism is poor. It generally needs multiple iterations to complete the current coding symbol (bin), and the coding of the subsequent coding symbol (bin) depends on the iteration result of the current coding symbol (bin). Moreover, in the process of normalization, it is necessary to continuously adjust the range of the encoding lower limit (low), and the calculation of normalization is very complicated. At present, some binary arithmetic coders can only achieve partial parallelization, and although some binary arithmetic coders can achieve complete parallelization, they cannot uniformly handle the three coding modes of conventional coding mode, equal probability coding mode and cut-off coding mode, namely The three coding modes are handled separately, and the reusability is poor. In addition, the current code stream extraction technology is relatively backward. Generally, multiple clock cycles are required to complete the code stream extraction operation of a coded symbol (bin). Therefore, it is necessary to rely on the method of adding data FIFO to solve the code stream extraction operation and normalization. time difference between operations.

Contents of the invention

The object of the present invention is to provide a binary arithmetic coder and its coding method, so as to solve the problem of poor parallelism in the existing binary arithmetic codes.

Another object of the present invention is to solve the problem of poor reusability of binary arithmetic coding in the prior art.

Another object of the present invention is to solve the problem of complex code stream extraction operations in the prior art.

In order to solve the above technical problems, the present invention provides a binary arithmetic coder for outputting code streams of binary arithmetic codes in multiple coding modes, including a coding interval update and normalization module, a coding lower limit update and normalization module and A code stream extraction module, the encoding interval update and normalization module, the encoding lower limit update and normalization module and the code stream extraction module form a parallel three-stage pipeline structure;

The coding interval update and normalization module, as a first-level pipeline engine, is used to extract the coding state, coding mode and coding symbols, update the coding interval and complete the normalization of the coding interval, and output the normalized coding interval The number of iterations and the coding interval to the coding lower limit update and normalization module;

The coding lower limit update and normalization module, as a second-level pipeline engine, is used to extract coding symbols, coding modes, the number of iterations of normalization of coding intervals and coding intervals, update the coding lower limit and complete the normalization of the coding lower limit , and output the overflow bit of the encoding lower limit and the number of iterations of the encoding lower limit normalization to the code stream extraction module;

The code stream extraction module, as a third-stage pipeline engine, is used to extract the number of iterations normalized by the encoding lower limit and the overflow bit of the encoding lower limit, and output a variable-length code stream and the length of the code stream.

Preferably, in the binary arithmetic coder, the encoding interval update and normalization module also outputs the interval length of the high-probability character to the encoding lower limit update and normalization module.

Preferably, in the binary arithmetic encoder, mode multiplexing is realized through a selector.

Preferably, in the binary arithmetic coder, the coding modes include normal coding mode, equal probability coding mode and cut-off coding mode.

Preferably, in the binary arithmetic encoder, when the mode is multiplexed, the encoding interval update and normalization module includes a first selector, a first subtractor, a second selector, a first 1 detection circuit, a first Shifter and the first D flip-flop; Wherein, the first selector is used to obtain the interval length of the low-probability character under the current encoding mode according to the encoding mode extracted; The interval length of the low-probability character in the current encoding mode is used to obtain the interval length of the high-probability character in the current encoding mode; the second selector is used to select from the current encoding interval and the current encoding mode according to the current encoding mode and the current encoding symbol Select one of the interval length of the small probability character and the interval length of the high probability character under the current encoding mode as the encoding interval before normalization to output; the first 1 detection circuit is used to encode the encoding interval before the normalization Count the leading 0s to obtain the number of iterations of the normalization of the coding interval; the first shifter is used to shift the coding interval before normalization to the left according to the number of iterations of the normalization of the coding interval and output the normalized After the normalized coding interval is passed through the first D flip-flop, the coding interval is output to be used in the next cycle.

Preferably, in the binary arithmetic encoder, when the encoding mode is an equal probability encoding mode, the encoding interval update and normalization module includes a first trigger, and the encoding interval is directly output as the normalized The coding interval, after the normalized coding interval passes through the first D flip-flop, outputs the coding interval to be used in the next cycle.

Preferably, in the binary arithmetic encoder, when the encoding mode is a cut-off encoding mode, the encoding interval update and normalization module includes a first subtractor, a second selector, a first 1 detection circuit, a first A shifter and the first D flip-flop; wherein, the first subtractor is used to obtain the interval length of the high-probability character under the current encoding mode according to the interval length of the low-probability character under the current encoding interval and the current encoding mode ; The second selector is used to select one of the interval lengths of low-probability characters in the current encoding mode or the interval length of high-probability characters in the current encoding mode as normalization according to the current encoding mode and the current encoding symbol The first 1 detection circuit is used to count the leading 0 of the coding interval before the normalization to obtain the number of iterations of the normalization of the coding interval; the first shifter is used for according to the The number of iterations of the normalization of the coding interval is to shift the coding interval before normalization to the left to output the normalized coding interval, and the normalized coding interval is passed through the first D flip-flop to output the coding interval and wait for the next cycle backup.

Preferably, in the binary arithmetic encoder, when the mode is multiplexed, the encoding lower limit update and normalization module includes a third selector, a fourth selector, a first adder, a fifth selector, a second shifter, a third shifter, a fourth shifter, a first comparator and a second D flip-flop; wherein, the third selector is used to select from the current Choose one of the encoding interval, the interval length of the high-probability character in the current encoding mode, and the default value of the encoding lower limit as the output; the fourth selector is used to select one of the encoding lower limit or 2 times the encoding lower limit according to the current encoding mode. output; the first adder is used to obtain the lower limit of coding before normalization according to the third selector and the fourth selector; the fifth selector is used to normalize from the coding interval according to the coding mode The number of iterations of the encoding lower limit normalization and the default value of the number of iterations of the encoding lower limit normalization are selected as the iteration number of the encoding lower limit normalization; the second shifter is used to normalize the number of iterations according to the encoding lower limit normalization The encoding lower limit before normalization is shifted to the left, and the lower 9 bits are taken to obtain the lower 9 bits of the encoding lower limit after normalization; the third shifter is used to shift the result obtained by the second shifter to the right by 9 bits to obtain the encoding The overflow bit of the lower limit; the fourth shifter is used to shift the constant 0x1ff to the right according to the number of iterations normalized by the lower limit of the encoding, and the result of the bitwise OR operation of the constant 0x1ff and the lower limit of the encoding before normalization is passed through the first The comparator compares with the constant 0x3ff, and obtains the highest bit of the normalized coding lower limit according to the comparison result. After the normalized coding lower limit passes through the second D flip-flop, the output coding lower limit is ready for use in the next cycle.

Preferably, in the binary arithmetic coder, when the current encoding mode is the regular encoding mode and the encoding symbol is the interval length of the low-probability character, the third selector selects the interval length of the high-probability character as output; the current encoding mode When it is an equal probability coding mode and the coding symbol is 1, the third selector selects the current coding interval as output; in other cases, the third selector selects the default value of the coding lower limit as output.

Preferably, in the binary arithmetic coder, when the current coding mode is an equal probability coding mode, the fourth selector selects twice the coding lower limit as the output, and otherwise selects the coding lower limit as the output.

Preferably, in the binary arithmetic coder, when the current coding mode is an equal probability coding mode, the fifth selector selects the default value of the number of iterations of coding lower limit normalization as output, and in other cases, the fifth selector selects The number of iterations normalized by the encoder to select the encoding interval is output.

Preferably, in the binary arithmetic coder, the fourth shifter shifts the constant 0x1ff to the right by the number of iterations of the normalized lower limit of the encoding, and then outputs the lower limit of the encoding before normalization by bitwise OR, and the fourth shifter A comparator compares the result of the bitwise OR output with the constant 0x3ff. If the result of the bitwise OR operation is equal to 0x3ff, the highest bit of the normalized coding lower limit is 1, otherwise it is 0.

Preferably, in the binary arithmetic coder, when the coding mode is a normal coding mode and/or a cut-off coding mode, the coding lower limit update and normalization module includes: a third selector, a first adder , the fifth selector, the second shifter, the third shifter, the fourth shifter, the first comparator and the second D flip-flop; wherein, the third selector is used to select according to the current encoding symbol The interval length of the high-probability character under the current encoding mode and the default value output of the encoding lower limit; the first adder is used to obtain the encoding lower limit before normalization according to the third selector encoding lower limit; the fifth selector uses According to the encoding mode, the number of iterations normalized in the encoding interval is selected as the number of iterations normalized by the lower limit of encoding to output; the second shifter is used to normalize the number of iterations normalized according to the lower limit of encoding The encoding lower limit before normalization is shifted to the left, and the lower 9 bits are taken to obtain the lower 9 bits of the encoding lower limit after normalization; the third shifter is used to shift the result obtained by the second shifter to the right by 9 bits to obtain the encoding The overflow bit of the lower limit; the fourth shifter is used to shift the constant 0x1ff to the right according to the number of iterations normalized by the lower limit of the encoding, and the result of the bitwise OR operation of the constant 0x1ff and the lower limit of the encoding before normalization is passed through the first The comparator compares with the constant 0x3ff, and obtains the highest bit of the normalized coding lower limit according to the comparison result. After the normalized coding lower limit passes through the second flip-flop, the output coding lower limit is ready for use in the next cycle.

Preferably, in the binary arithmetic coder, when the coding mode is an equal-probability coding mode, the coding lower limit update and normalization module includes a third selector, a first adder, a fifth selector, The second shifter, the third shifter, the fourth shifter, the first comparator, and the second D flip-flop; wherein, the third selector is used to select the current encoding interval according to the current encoding symbol and the default value of the encoding lower limit as output; the first adder is used to obtain the encoding lower limit before normalization according to the third selector and 2 times the encoding lower limit; the fifth selector is used to obtain the encoding lower limit according to the encoding mode , select the default value of the number of iterations of normalization of the lower limit of encoding as the number of iterations of normalization of the lower limit of encoding to output; the second shifter is used to convert the lower limit of encoding before normalization according to the number of iterations of normalization of the lower limit of encoding Shifting left, taking the lower 9 bits to obtain the lower 9 bits of the normalized encoding lower limit; the third shifter is used to shift the result obtained by the second shifter to the right by 9 bits to obtain the overflow bit of the encoding lower limit; The fourth shifter is used to shift the constant 0x1ff to the right according to the number of iterations normalized by the lower limit of the encoding. After the constant 0x1ff is shifted to the right, the result of the bitwise OR operation with the lower limit of the encoding before normalization is passed through the first comparator and the constant 0x3ff The comparison is performed, and the highest bit of the normalized coding lower limit is obtained according to the comparison result. After the normalized coding lower limit passes through the second D flip-flop, the coding lower limit is output for standby in the next cycle.

Preferably, in the binary arithmetic encoder, the code stream extraction module includes a first 0 detection circuit, a sixth selector, a third D flip-flop, a seventh selector, and a code stream output unit; wherein, the first 0 detection The circuit is used to detect the first 0 at the end of the overflow bit of the lower limit of the encoding, and output the number of 1s that appear after the first 0, the number of 1s that appear after the first 0 and the iteration of the normalization of the lower limit of the encoding The number of times to compare to obtain the comparison result; the sixth selector is used for the number of 1s that appear after the first 0, the default value of the number of uncertain bits and the number of iterations normalized by the encoding lower limit according to the comparison result Select one of the sum of the number of uncertain bits and output it as the number of uncertain bits; the third D flip-flop is used to store the number of uncertain bits output by the sixth selector as a backup for the next cycle; The seventh selector is used to select from the sum of the number of iterations normalized by the lower limit of encoding and the number of uncertain bits, the default value of the code stream length, and the number of iterations and uncertainties of normalization of the lower limit of encoding according to the comparison result. Choose one of the difference between the sum of the number of bits and the number of 1s that appear after the first 0 as the code stream length and output it; the code stream output unit is used to output a variable-length code stream according to the overflow bit of the encoding lower limit .

Preferably, in the binary arithmetic encoder, when the number of iterations of the encoding lower limit normalization is less than the number of 1s that appear after the first 0, the comparison result is 0; when the encoding lower limit normalization When the number of iterations of normalization is equal to the number of 1s that appear after the first 0, the comparison result is 1; when the number of iterations of the encoding lower limit normalization is greater than the number of 1s that appear after the first 0 When counting, the comparison result is 2.

Preferably, in the binary arithmetic encoder, when the comparison result is 0, the sixth selector outputs the sum of the number of iterations normalized by the encoding lower limit and the number of uncertain bits as output; when When the comparison result is 1, the sixth selector outputs the default value of the number of uncertain bits; when the comparison result is 2, the sixth selector outputs the first 0 The number of 1s that appear after that is output.

Preferably, in the binary arithmetic coder, when the comparison result is 0, the length of the code stream output by the seventh selector is the difference between the number of iterations normalized by the encoding lower limit and the number of uncertain bits. and; when the comparison result is 1, the code stream length output by the seventh selector is the default value of the code stream length; when the comparison result is 2, the code stream output by the seventh selector The stream length is the difference between the sum of the number of iterations normalized by the encoding lower limit and the number of uncertain bits, and the number of 1s that appear after the first 0.

Preferably, in the binary arithmetic encoder, the code stream output unit includes an eighth selector, a fifth shifter and a sixth shifter; when the highest bit of the effective code stream is 0, the eighth The selector outputs the first constant, and when the highest bit of the effective code stream is 1, the eighth selector outputs the second constant; the fifth shifter shifts the eighth selector’s The output result is shifted to the right, and the sixth shifter shifts the output result of the fifth shifter to the left according to the sum of the number of iterations normalized by the encoding lower limit and 1, and finally outputs it with the last bit by bitwise OR.

Preferably, in the binary arithmetic encoder, the code stream output unit sequentially outputs a highest bit, several middle bits and several last bits in a left-aligned manner; wherein, the highest bit is the middle bit when 1 All are 0, when the highest bit is 0, the middle bits are all 1, and the number of the middle bits is equal to the number of uncertain bits.

Preferably, in the binary arithmetic encoder, the code stream extraction module completes the code stream extraction operation within one clock cycle.

Correspondingly, the present invention also provides a coding interval update and normalization module, which is used to extract the coding state, coding mode and coding symbol, update the coding interval and complete the normalization of the coding interval, and output the normalized coding interval The number of iterations and the encoding interval;

The encoding interval update and normalization module includes a first selector, a first subtractor, a second selector, a first 1 detection circuit, a first shifter and a first D flip-flop; wherein,

The first selector is used to obtain the interval length of the low-probability character in the current encoding mode according to the extracted encoding mode;

The first subtractor is used to obtain the interval length of the high-probability character under the current encoding mode according to the interval length of the current encoding interval and the low-probability character under the current encoding mode;

The second selector is used to select one of the current encoding interval, the interval length of the low-probability character in the current encoding mode, and the interval length of the high-probability character in the current encoding mode as a normalization according to the current encoding mode and the current encoding symbol Output the coding interval before conversion;

The first 1 detection circuit is used to count the leading 0s of the coding interval before the normalization to obtain the number of iterations of the normalization of the coding interval;

The first shifter is used to left-shift the coding interval before normalization according to the number of iterations of the normalization of the coding interval to output the normalized coding interval, and the normalized coding interval passes through the second After a D flip-flop, the output coding interval is reserved for the next cycle.

Preferably, in the encoding interval update and normalization module, when the current encoding mode is a conventional encoding mode or an equal probability encoding mode, look up the table to obtain the initial value of the interval length of 4 low probability characters, and the first selection The device selects one of the initial values of the interval lengths of four low-probability characters as the interval length of the low-probability characters in the current encoding mode; when the current encoding mode is the cut-off encoding mode, the interval length of the low-probability characters is equal to 2.

Preferably, in the coding interval update and normalization module, when the current coding mode is an equal-probability coding mode, the second selector selects the current coding range as an output; the current coding mode is a cut-off coding mode and the current When the coding symbol is 1, or when the current coding mode is in the conventional coding mode and the current coding symbol is 0, the second selector selects the interval length of the low-probability character in the current coding mode as output; in other cases, the second selection The device selects the interval length of the high-probability character in the current encoding mode as the output.

Correspondingly, the present invention also provides a coding lower limit update and normalization module, which is used to extract coding symbols, coding modes, the number of iterations of normalization of coding intervals and coding intervals, update the coding lower limit and complete the normalization of the coding lower limit , and output the overflow bit of the encoding lower limit and the number of iterations of the encoding lower limit normalization;

The encoding lower limit update and normalization module includes a third selector, a fourth selector, a first adder, a fifth selector, a second shifter, a third shifter, a fourth shifter, a a comparator and a second D flip-flop; where,

The third selector is used to select one of the current encoding interval, the interval length of the high probability character in the current encoding mode, and the default value of the encoding lower limit as output according to the current encoding mode and the current encoding symbol;

The fourth selector is used to select one of the encoding lower limit or 2 times the encoding lower limit as output according to the current encoding mode;

The first adder is used to obtain the coding lower limit before normalization according to the third selector and the fourth selector;

The fifth selector is used to select one of the default values of the number of iterations normalized by the encoding interval and the number of iterations normalized by the lower limit of encoding according to the encoding mode to output as the number of iterations normalized by the lower limit of encoding;

The second shifter is used to shift the lower limit of encoding before normalization to the left according to the number of iterations of the lower limit of encoding normalization, and take the lower 9 bits to obtain the lower 9 bits of the lower limit of encoding after normalization;

The third shifter is used to shift the result obtained by the second shifter to the right by 9 bits to obtain the overflow bit of the lower limit of encoding;

The fourth shifter is used to shift the constant 0x1ff to the right according to the number of iterations normalized by the lower limit of the encoding. After the constant 0x1ff is shifted to the right, the result of the bitwise OR operation with the lower limit of the encoding before normalization is passed through the first comparator and the constant 0x3ff The comparison is performed, and the highest bit of the normalized coding lower limit is obtained according to the comparison result. After the normalized coding lower limit passes through the second D flip-flop, the coding lower limit is output for standby in the next cycle.

Preferably, in the described encoding lower limit update and normalization module, when the current encoding mode is the conventional encoding mode and the encoding symbol is the interval length of the low-probability character, the interval length of the high-probability character selected by the third selector is Output; when the current encoding mode is an equal-probability encoding mode and the encoding symbol is 1, the third selector selects the current encoding interval as output; in other cases, the third selector selects the default value of the encoding lower limit as output.

Preferably, in the updating and normalization module of the coding lower limit, when the current coding mode is an equal-probability coding mode, the fourth selector selects twice the coding lower limit as the output, and otherwise selects the coding lower limit as the output.

Preferably, in the updating and normalization module of the lower limit of encoding, when the current encoding mode is an equal-probability encoding mode, the fifth selector selects the default value of the number of iterations of the lower limit of encoding normalization as output, and in other cases The fifth selector selects the number of iterations of encoding interval normalization as an output.

Preferably, in the updating and normalization module of the encoding lower limit, the fourth shifter shifts the constant 0x1ff to the right by the number of iterations of the encoding lower limit normalization and bitwise OR with the encoding lower limit before normalization Output, the first comparator compares the result of the bitwise OR output with the constant 0x3ff, if the result of the bitwise OR operation is equal to 0x3ff, the highest bit of the normalized coding lower limit is 1, otherwise it is 0.

Correspondingly, the present invention also provides a code stream extraction module, which is used to extract the number of iterations of normalization of the encoding lower limit and the overflow bit of the encoding lower limit, and output a variable-length code stream and the length of the code stream;

The code stream extraction module includes a first 0 detection circuit, a sixth selector, a third D flip-flop, a seventh selector and a code stream output unit; wherein,

The first 0 detection circuit is used to detect the first 0 at the end of the overflow bit of the lower limit of the encoding, and output the number of 1s that appear after the first 0, and the number of 1s that appear after the first 0 is normalized to the lower limit of the encoding The number of iterations optimized is compared to obtain the comparison result;

The sixth selector is used for the sum of the number of 1s that appear after the first 0, the default value of the number of uncertain bits, and the number of iterations normalized by the encoding lower limit and the number of uncertain bits according to the comparison result Choose one of them as the number of uncertain bits to output;

The third D flip-flop is used to store the number of uncertain bits output by the sixth selector as a backup for the next cycle;

The seventh selector is used to select from the sum of the number of iterations normalized by the lower limit of encoding and the number of uncertain bits, the default value of the code stream length, and the number of iterations and uncertainties of normalization of the lower limit of encoding according to the comparison result. Choose one of the difference between the sum of the number of bits and the number of 1s that appear after the first 0 as the code stream length to output;

A code stream output unit, configured to output a variable-length code stream according to the overflow bits of the encoding lower limit.

Preferably, in the code stream extraction module, when the number of iterations of normalization of the encoding lower limit is less than the number of 1s that appear after the first 0, the comparison result is 0; when the encoding lower limit When the number of iterations of normalization is equal to the number of 1s that appear after the first 0, the comparison result is 1; when the number of iterations of the encoding lower limit normalization is greater than the number of 1s that appear after the first 0 number, the comparison result is 2.

Preferably, in the code stream extraction module, when the comparison result is 0, the sixth selector outputs the sum of the number of iterations normalized by the encoding lower limit and the number of uncertain bits; When the comparison result is 1, the sixth selector outputs the default value of the number of uncertain bits; when the comparison result is 2, the sixth selector outputs the first The number of 1s that appear after 0 is output.

Preferably, in the code stream extraction module, when the comparison result is 0, the length of the code stream output by the seventh selector is the number of iterations and the number of uncertain bits normalized by the encoding lower limit sum; when the comparison result is 1, the code stream length output by the seventh selector is the default value of the code stream length; when the comparison result is 2, the code stream length output by the seventh selector The code stream length is the difference between the sum of the number of iterations normalized by the encoding lower limit and the number of uncertain bits and the number of 1s that appear after the first 0.

Preferably, in the code stream extraction module, the code stream output unit includes an eighth selector, a fifth shifter and a sixth shifter; when the highest bit of the effective code stream is 0, the first The eighth selector outputs the first constant. When the highest bit of the effective code stream is 1, the eighth selector outputs the second constant; the fifth shifter shifts the eighth selector to The output result of the fifth shifter is shifted to the right, and the sixth shifter shifts the output result of the fifth shifter to the left according to the sum of the number of iterations normalized by the encoding lower limit and 1, and finally outputs it with the last bit.

Preferably, in the code stream extraction module, the code stream output unit sequentially outputs a highest bit, several middle bits, and several last bits in a left-aligned manner; wherein, the highest bit is 1 when the middle The bits are all 0, and when the highest bit is 0, the middle bits are all 1, and the number of the middle bits is equal to the number of uncertain bits.

Preferably, in the code stream extraction module, the code stream extraction module completes the code stream extraction operation within one clock cycle.

Correspondingly, the present invention also provides a coding method of a binary arithmetic coder, and the coding method of the binary arithmetic coder includes:

Step S10: The encoding interval update and normalization module acts as the first-level pipeline engine to extract the encoding state, encoding mode and encoding symbols, update the encoding interval and complete the normalization of the encoding interval, and output the number of iterations of the normalization of the encoding interval and the coding interval;

Step S11: The coding lower limit update and normalization module is used as the second-level pipeline engine to extract the coding symbol, coding mode, the number of iterations of the normalization of the coding interval and the coding interval, update the coding lower limit and complete the normalization of the coding lower limit, and Output the overflow bit of the encoding lower limit and the number of iterations of the encoding lower limit normalization;

Step S12: The code stream extraction module acts as a third-stage pipeline engine, extracts the number of iterations normalized by the encoding lower limit and overflow bits of the encoding lower limit, and outputs a variable-length code stream and the length of the code stream.

Preferably, in the encoding method of the binary arithmetic coder, in step S10, the encoding interval update and normalization module also outputs the interval length of the high-probability character to the encoding lower limit update and normalization module.

Preferably, in the coding method of the binary arithmetic coder, the coding modes include normal coding mode, equal probability coding mode and cut-off coding mode.

Preferably, in the encoding method of the binary arithmetic encoder, the step S10 further includes:

The first selector obtains the interval length of the low-probability character in the current encoding mode according to the encoding mode;

The first subtractor obtains the interval length of the high-probability character under the current encoding mode according to the interval length of the current encoding interval and the low-probability character under the current encoding mode;

According to the current encoding mode and the current encoding symbol, the second selector selects one of the current encoding interval, the interval length of the low-probability character in the current encoding mode, or the interval length of the high-probability character in the current encoding mode as normalization Output the coding interval before conversion;

The first 1 detection circuit carries out leading 0 counting to the coding interval before the normalization, and obtains the number of iterations of the normalization of the coding interval;

The first shifter performs left shift on the coding interval before normalization according to the number of iterations of the normalization of the coding interval to output the normalized coding interval, and the normalized coding interval is passed through a D flip-flop , the output coding interval is reserved for the next cycle.

Preferably, the coding method of the binary arithmetic coder is characterized in that, when the current coding mode is the conventional coding mode or the equal probability coding mode, the initial value of the interval length of 4 characters with low probability is obtained by looking up the table, and the first The selector selects one of the initial interval lengths of four low-probability characters as the interval length of low-probability characters in the current encoding mode; when the current encoding mode is the cut-off encoding mode, the interval length of the low-probability characters is equal to 2.

Preferably, in the coding method of the binary arithmetic coder, when the current coding mode is an equal-probability coding mode, the second selector selects the current coding interval as an output; the current coding mode is a cut-off coding mode and the current coding symbol is 1, or when the current encoding mode is in the conventional encoding mode and the current encoding symbol is 0, the second selector selects the interval length of the low-probability character in the current encoding mode as output; in other cases, the second selector selects The interval length of the high-probability characters in the current encoding mode is output.

Preferably, the coding method of the binary arithmetic coder is characterized in that step S11 further includes:

According to the current encoding mode and the current encoding symbol, the third selector selects one of the current encoding interval, the interval length of the high probability character in the current encoding mode or the default value of the encoding lower limit as an output;

The fourth selector selects one of the encoding lower limit or 2 times the encoding lower limit as output according to the current encoding mode;

The first adder obtains the coding lower limit before normalization according to the third selector and the fourth selector;

According to the current coding mode, the fifth selector selects one of the default values of the number of iterations normalized by the coding interval and the number of iterations normalized by the lower limit of coding to output as the number of iterations of the lower limit normalization of coding;

The second shifter shifts the encoding lower limit before normalization to the left according to the number of iterations of the encoding lower limit normalization, and takes the lower 9 bits to obtain the lower 9 bits of the normalized encoding lower limit;

The third shifter shifts the result obtained by the second shifter to the right by 9 bits to obtain the overflow bit of the lower limit of encoding;

The fourth shifter shifts the constant 0x1ff to the right according to the number of iterations normalized by the lower limit of the encoding, and the result of the bitwise OR operation of the constant 0x1ff after the right shift and the lower limit of the encoding before normalization is compared with the constant 0x3ff by the first comparator, According to the comparison result, the highest bit of the normalized coding lower limit is obtained;

After the normalized encoding lower limit passes through the third D flip-flop, the output encoding lower limit is ready for use in the next cycle.

Preferably, in the encoding method of the binary arithmetic coder, when the current encoding mode is the normal encoding mode and the encoding symbol is the interval length of the low-probability character, the third selector selects the interval length of the high-probability character as output; When the current encoding mode is an equal-probability encoding mode and the encoding symbol is 1, the third selector selects the current encoding interval as output; in other cases, the third selector selects the default value of the encoding lower limit as output.

Preferably, in the coding method of the binary arithmetic coder, when the current coding mode is an equal-probability coding mode, the fourth selector selects twice the coding lower limit as the output, and otherwise selects the coding lower limit as the output.

Preferably, in the encoding method of the binary arithmetic encoder, when the current encoding mode is an equal probability encoding mode, the default value of the number of iterations selected by the fifth selector for encoding lower limit normalization is output, and the other conditions The fifth selector selects the number of iterations of encoding interval normalization as output.

Preferably, in the encoding method of the binary arithmetic encoder, the fourth shifter shifts the constant 0x1ff to the right by the number of iterations of the normalized lower limit of the encoding, and outputs the bitwise OR output with the lower limit of the encoding before normalization, The first comparator compares the result of the bitwise OR output with the constant 0x3ff. If the result of the bitwise OR operation is equal to 0x3ff, the highest bit of the normalized coding lower limit is 1, otherwise it is 0.

Preferably, in the encoding method of the binary arithmetic encoder, step S12 further includes:

The first 0 detection circuit detects the first 0 at the end of the overflow bit of the lower limit of the encoding, and outputs the number of 1s that appear after the first 0, the number of 1s that appear after the first 0 and the iteration of the normalization of the lower limit of the encoding times to compare;

The sixth selector selects one from the number of 1s that appear after the first 0, the default value of the number of uncertain bits, and the sum of the number of iterations normalized by the encoding lower limit and the number of uncertain bits according to the comparison result Output as an indeterminate number of bits;

The third D flip-flop stores the number of uncertain bits output by the sixth selector;

The seventh selector selects from the sum of the number of iterations normalized by the lower limit of encoding and the number of uncertain bits, the default value of the code stream length, and the number of iterations and the number of uncertain bits normalized by the lower limit of encoding according to the comparison result The difference between the sum and the number of 1s that appear after the first 0 is selected as the code stream length for output;

The code stream output unit outputs a variable-length code stream according to the overflow bits of the encoding lower limit.

Preferably, in the encoding method of the binary arithmetic encoder, when the number of iterations of the encoding lower limit normalization is less than the number of 1s that appear after the first 0, the comparison result is 0; when the When the number of iterations of the encoding lower limit normalization is equal to the number of 1s that appear after the first 0, the comparison result is 1; when the number of iterations of the encoding lower limit normalization is greater than the number of 1s that appear after the first 0 1, the comparison result is 2.

Preferably, in the encoding method of the binary arithmetic encoder, when the comparison result is 0, the sixth selector uses the sum of the number of iterations normalized by the lower limit of encoding and the number of uncertain bits as output; when the comparison result is 1, the sixth selector takes the default value of the number of uncertain bits as output; when the comparison result is 2, the sixth selector takes the first The number of 1s following a 0 is output.

Preferably, in the encoding method of the binary arithmetic encoder, when the comparison result is 0, the length of the code stream output by the seventh selector is equal to the number of iterations normalized by the encoding lower limit and the uncertain bit The sum of numbers; when the comparison result is 1, the code stream length output by the seventh selector is the default value of the code stream length; when the comparison result is 2, the seventh selector The length of the output code stream is the difference between the sum of the number of iterations normalized by the encoding lower limit and the number of uncertain bits and the number of 1s appearing after the first 0.

Preferably, in the encoding method of the binary arithmetic encoder, when the highest bit of the effective code stream is 0, the eighth selector outputs the first constant; when the highest bit of the effective code stream is 1, the eighth selector The selector outputs the second constant; the fifth shifter shifts the output of the eighth selector to the right according to the difference between 63 and the number of uncertain bits, and the sixth shifter normalizes the number of iterations according to the encoding lower limit The sum of 1 and the output result of the fifth shifter is shifted to the left, and finally output with the bitwise OR of the last bit.

Preferably, in the encoding method of the binary arithmetic encoder, the code stream output unit sequentially outputs a highest bit, several middle bits and several last bits in a left-aligned manner; wherein, the highest bit is 1 When the middle bits are all 0, when the highest bit is 0, the middle bits are all 1, and the number of the middle bits is equal to the number of uncertain bits.

Preferably, in the coding method of the binary arithmetic coder, the code stream extraction module completes the code stream extraction operation within one clock cycle.

In the binary arithmetic encoder and its encoding method provided by the present invention, a three-stage pipeline structure is formed by the encoding interval update and normalization module, the encoding lower limit update and normalization module, so that the code stream extraction module can Internal processing of one or more coding symbols realizes complete parallelism and solves the problem of poor parallelism in the prior art; in addition, the sharing of different coding modes is realized through the selector, which improves the reusability of the circuit and reduces the hardware cost; In addition, the code stream extraction operation of the code stream extraction method is simple and does not require additional FIFO resources.

Description of drawings

Fig. 1 is the functional block diagram of the binary arithmetic coder of the embodiment of the present invention;

FIG. 2 is a schematic diagram of the workflow of the encoding interval update and normalization module under the mode multiplexing of the embodiment of the present invention;

FIG. 3 is a schematic diagram of the workflow of the encoding interval update and normalization module in the conventional encoding mode of the embodiment of the present invention;

4 is a schematic diagram of the workflow of the coding interval update and normalization module in the equal probability coding mode of the embodiment of the present invention;

5 is a schematic diagram of the workflow of the coding interval update and normalization module in the cut-off coding mode according to the embodiment of the present invention;

A schematic diagram of the workflow of the encoding lower limit update and normalization module under the mode multiplexing of the embodiment of the present invention in FIG. 6 ;

The schematic diagram of the workflow of the normal coding mode and the coding lower limit update and normalization module in the cut-off coding mode of the embodiment of the present invention in FIG. 7 ;

In Fig. 8, a schematic diagram of the workflow of the encoding lower limit update and normalization module in the equal probability encoding mode of the embodiment of the present invention;

FIG. 9 is a schematic structural diagram of a code stream extraction module according to an embodiment of the present invention;

FIG. 10 is a flowchart of a coding method of a binary arithmetic coder according to an embodiment of the present invention;

FIG. 11 is a schematic flowchart of step S10 in FIG. 10;

FIG. 12 is a schematic flowchart of step S11 in FIG. 10;

FIG. 13 is a schematic flow chart of step S12 in FIG. 10 .

detailed description

The binary arithmetic encoder and its encoding method proposed by the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

Please refer to FIG. 1 , which is a functional block diagram of a binary arithmetic encoder according to an embodiment of the present invention. The binary arithmetic coder is used to output binary arithmetic coded code streams of various encoding modes, as shown in FIG. The normalization module 20 and the code stream extraction module 30, the encoding interval update and normalization module 10, the encoding lower limit update and normalization module 20 and the code stream extraction module 30 form a parallel three-stage pipeline structure.

Coding interval update and normalization module 10, as the first-level pipeline engine, is used to extract the encoding state state, encoding mode mode and encoding symbol Bin, update the encoding interval range and complete the normalization of the encoding interval range, and output the encoding interval The normalized number of iterations S, the encoding interval range and the interval length rMPS of high-probability characters are sent to the next-level pipeline engine, which is the encoding lower limit update and normalization module 20 .

The encoding lower limit update and normalization module 20, as a second-level pipeline engine, is used to extract the encoding symbol Bin, encoding mode mode, the number of iterations S of encoding interval normalization, encoding interval range, and the interval length rMPS of high-probability characters, Update the encoding lower limit (low) and complete the normalization of the encoding lower limit (low), and output the overflow bit overflow_low of the encoding lower limit and the iteration number S1 of the encoding lower limit normalization to the next-level pipeline engine, that is, the stream extraction module 30.

The code stream extracting module 30, as a third-level pipeline engine, is used to extract the iteration number S1 of normalization of the encoding lower limit and the overflow bit overflow_low of the encoding lower limit, and output a variable-length code stream bit_stream and a code stream length bit_len.

The present invention adopts a three-stage pipeline structure, that is, adopts a highly parallel full pipeline design, without adding any control logic circuit, and can process one or more coding symbols in one clock cycle, thereby realizing complete parallelism and solving the problems in the prior art. Parallelism problem.

Specifically, the coding mode mode includes a normal coding mode, an equal-probability coding mode and a cut-off coding mode. In this embodiment, mode is 0, which means the normal coding mode, mode is 1, which means the equal probability coding mode, and mode is 2, which means the cut-off coding mode.

In this embodiment, the encoding interval update and normalization module 10 and the encoding lower limit update and normalization module 20 in the binary arithmetic encoder realize the sharing of different encoding modes through the selector, which improves the reusability of the circuit, Increased circuit throughput rate and reduced hardware overhead.

Each module will be described in detail below.

Fig. 2 is a schematic diagram of the workflow of the encoding interval update and normalization module 10 under the mode multiplexing of the embodiment of the present invention, and Fig. 3 is a schematic diagram of the encoding interval update and normalization module 10 under the conventional encoding mode of the embodiment of the present invention Schematic diagram of the workflow, FIG. 4 is a schematic diagram of the workflow of the coding interval update and normalization module 10 in the equal-probability coding mode of the embodiment of the present invention, and FIG. A schematic diagram of the workflow of the chemical module 10.

The structure and working process of the encoding interval update and normalization module 10 under mode multiplexing will be described in detail below in conjunction with FIG. 2 . As shown in FIG. 2 , the encoding interval update and normalization module 10 under mode multiplexing includes a first selector 11, a second selector 12, a first subtractor 13, a first 1 detection circuit 15, and a first shifter 16 and the first D flip-flop 17 .

Wherein, the first selector 11 is used to obtain the interval length rLPS of the low-probability character in the current encoding mode according to the extracted encoding mode mode, and output the interval length rLPS of the low-probability character in the current encoding mode; specifically, When the encoding mode mode is 0 or 1, that is, in the conventional encoding mode or the equal probability encoding mode, the interval length rLPS of the low probability character is obtained according to the probability state state table lookup result and range[7:6], as a non-restrictive For example, the probability state state look-up table obtains 4 rLPS initial values (rLPS0, rLPS1, rLPS2, rLPS3), and the first selector 11 selects one of the 4 initial values according to range[7:6] as the current encoding mode (standard mode ) under the low-probability character interval length rLPS output;

Wherein, the first subtractor 13 is used to obtain the interval length rMPS of the high-probability character under the current encoding mode according to the interval length rLPS of the low-probability character under the current encoding interval range and the current encoding mode, and the calculation formula of rMPS is: rMPS=range-rLPS.

Wherein, the second selector 12 is used to select from the current coding interval range, the interval length rLPS of the low-probability character in the current coding mode and the high-probability character in the current coding mode according to the current coding mode mode and the current coding symbol bin. Select one of the interval length rMPS as the encoding interval range _nor before normalization for output. Specifically, in the case of mode multiplexing, the input of the second selector 12 is the interval length rLPS of the low-probability character in the current encoding mode, the interval length rMPS of the high-probability character in the current encoding mode, and the current encoding interval range, select The factor is the current encoding mode mode and the current encoding symbol bin; if the current encoding mode mode is 1, select the current encoding interval range as output; if the current encoding mode mode is 2 and the current encoding symbol bin is 1, or the current encoding mode mode is 0 and the current encoding symbol bin is 0, select the interval length rLPS of the low-probability character in the current encoding mode as the output; in other cases except the above two, select the interval length rMPS of the high-probability character in the current encoding mode as output.

Among them, the first 1 detection circuit 15 is used to count the leading 0s of the coding interval range _nor before normalization to obtain the iteration number S of the normalization of the coding interval, and the iteration number S of the normalization of the coding interval will be input to the next Level pipeline engine, that is, the encoding lower limit update and normalization module 20.

Wherein, the first shifter 16 is used to obtain the normalized coding range rangeO according to the normalized coding range range _nor and the number of iterations S of coding range normalization, that is, by The encoding interval range _nor shifts left by S bits to obtain the normalized encoding interval range0; after the normalized encoding interval rangeO passes through the first D flip-flop 17, the output encoding interval range is to be updated with the normalization module 10 Reserved for the next cycle.

The structure and working process of the coding range update and normalization module 10 in the conventional coding mode will be described in detail below in conjunction with FIG. 3 . As shown in FIG. 3 , the coding interval update and normalization module 10 in the conventional coding mode includes a first subtractor 13, a second selector 12, a first 1 detection circuit 15, a first shifter 16 and a first D flip-flop 17.

Wherein, the first subtractor 13 is used to obtain the interval length rMPS of the high-probability character under the current encoding mode according to the interval length rLPS of the low-probability character of the current encoding interval range and the current encoding mode mode; the calculation formula of rMPS is: rMPS=range-rLPS. As a non-limiting example, the interval length rLPS of the low-probability character is obtained according to the probability state state table lookup result and range[7:6]. Probability state state lookup table to get 4 rLPS initial values (rLPS0, rLPS1, rLPS2, rLPS3), select one of the 4 rLPS initial values according to range[7:6] as the interval length rLPS of small probability characters and output to the first subtraction The first subtractor 13 obtains the interval length rMPS of the high-probability character in the current encoding mode according to the current encoding interval range and the interval length rLPS of the low-probability character in the current encoding mode mode.

Wherein, the second selector 12 is used to select one of the interval length rLPS of the low-probability character in the current encoding mode and the interval length rMPS of the high-probability character in the current mode according to the current encoding symbol bin as the pre-normalization The encoding interval range _nor is output.

Among them, the first 1 detection circuit 15 is used to count the leading 0s of the coding interval range _nor before normalization to obtain the iteration number S of the normalization of the coding interval, and the iteration number S of the normalization of the coding interval will be input to the next Level pipeline engine, that is, the encoding lower limit update and normalization module 20.

Wherein, the first shifter 16 is used to obtain the normalized coding range rangeO according to the normalized coding range range _nor and the number of iterations S of coding range normalization, that is, by The coding interval range _nor shifted to the left by S bits to obtain the normalized coding interval range0, after the normalized coding interval rangeO passes through the first D flip-flop 17, the output coding interval range is to be updated with the normalization module 10 Reserved for the next cycle.

The structure and working process of the coding interval update and normalization module 10 in the equal-probability coding mode will be described in detail below in conjunction with FIG. 4 . As shown in FIG. 4 , the coding range update and normalization module 10 in the equal probability coding mode includes a first D flip-flop 17 , and the current coding range range is directly output as a normalized coding range rangeO. After the normalized coding interval range0 passes through the first D flip-flop 17, the output coding interval range is to be updated with the normalization module 10 for standby in the next cycle.

The structure and working process of the coding interval update and normalization module 10 in the cut-off coding mode will be described in detail below in conjunction with FIG. 5 . As shown in FIG. 5 , the coding interval update and normalization module 10 in the cut-off coding mode includes: a first subtractor 13, a second selector 12, a first 1 detection circuit 15, a first shifter 16 and a first shifter 16. A D flip-flop 17 .

Wherein, the first subtractor 13 is used to obtain the current encoding mode according to the interval length rLPS of the low-probability character of the current encoding interval range and the current encoding mode mode (under the cut-off encoding mode, the interval length rLPS of the low-probability character is equal to 2) The interval length rMPS of high probability characters in mode; the calculation formula of rMPS is: rMPS=range-rLPS.

Wherein, the second selector 12 is used to select one of the interval length rLPS of the low probability character in the current encoding mode mode or the interval length rMPS of the high probability character in the current mode mode according to the current encoding mode mode and the current encoding symbol bin It is output as the encoding interval range _nor before normalization.

Wherein, the first 1 detection circuit 15 is used for counting the leading 0s of the encoding interval range _nor before normalization, so as to obtain the iteration number S of the encoding interval normalization. The number of iterations S of the normalization of the encoding interval will be input to the next-level pipeline engine, that is, the encoding lower limit update and normalization module 20 .

Wherein, the first shifter 16 is used to obtain the normalized coding range rangeO according to the normalized coding range range _nor and the number of iterations S of coding range normalization, that is, by The coding interval range _nor is shifted to the left by S bits to obtain the normalized coding interval rangeO. After the normalized coding interval range0 passes through the first D flip-flop 17, the output coding interval range is to be updated with the normalization module 10 for standby in the next cycle.

Schematic diagram of the workflow of the coding lower limit update and normalization module under the mode multiplexing of the embodiment of the present invention in FIG. Schematic diagram of the workflow of the module, and a schematic diagram of the workflow of the lower limit update and normalization module of the embodiment of the present invention in the equal-probability coding mode in FIG. 8 .

The structure and working process of the encoding lower limit update and normalization module 20 under mode multiplexing in the embodiment of the present invention will be described in detail below in conjunction with FIG. 6 . As shown in Figure 6, the encoding lower limit update and normalization module 20 under the mode multiplexing includes a third selector 21, a fourth selector 22, a first adder 23, a fifth selector 24, a second shifter 25, a third shifter 26, a fourth shifter 27, a first comparator 28 and a second D flip-flop 29.

Wherein, the interval length rMPS of the input high-probability character of the third selector 21 and the default value of the current coding interval range and a coding lower limit (can be 0 here), the selection factor is the coding mode mode and the coding symbol bin, and the output is low_add1. Specifically, when the encoding mode mode is 0 and the encoding symbol bin is the interval length rLPS of a small probability character, the low_add1 output by the third selector 21 is equal to rMPS; when the mode is 1 and the bin is 1, the output low_add1 is equal to range ; otherwise output low_add1 is equal to 0.

Wherein, the input of the fourth selector 22 is the encoding lower limit low or twice the encoding lower limit (ie 2*low), the selection factor is the encoding mode mode, and the output is low_add0. Specifically, when the mode is 1, the low_add0 output by the fourth selector 22 is 2*low, and low is selected as the output in other cases.

Wherein, the first adder 23 obtains the encoding lower limit low _nor before normalization according to the outputs of the third selector 21 and the fourth selector 22 .

Wherein, the fifth selector 24 obtains the iteration number S1 of encoding lower limit normalization according to the encoding mode mode. The input of the 5th selector 24 is the default value (here can be 1) of the number of iterations S of normalized encoding interval normalization and the number of iterations of encoding lower limit normalization, the selection factor is the encoding mode mode, and the output is normalized by the lower limit of encoding The number of iterations S1 of optimization. Specifically, when the mode is 1, that is, in the equal-probability coding mode, the number of iterations S1 of the encoding lower limit normalization output by the fifth selector 24 is equal to 1; in other modes, the number of iterations S1 of the encoding lower limit normalization is equal to S.

Wherein, the second shifter 25 shifts the coding lower limit low _nor before normalization to the left by S1 bits, and takes the lower 9 bits to obtain the lower 9 bits of the normalized coding lower limit low0, and the second shifter 25 The obtained result is shifted right by 9 bits by the third shifter 26 to obtain the overflow bit overflow_low of the encoding lower limit. The overflow bit overflow_low of the encoding lower limit is input to the next-level pipeline engine, that is, the code stream extraction module 30 . In this embodiment, the number of overflow bits overflow_low of the encoding lower limit is 8 bits, but it should be understood that the present invention does not limit the number of overflow bits overflow_low of the encoding lower limit.

Among them, the highest bit of the normalized coding lower limit lowO is obtained by the following method: judging whether the 10th bit of low _nor and the subsequent S1 bits are all 1, if they are all 1, then the normalized The highest bit of the encoding lower limit lowO is 1, otherwise it is 0. The specific implementation is as follows, the fourth shifter 27 shifts a constant 0x1ff to the right by S1 bits and outputs the code lower limit low _nor before normalization, and the first comparator 28 outputs the result of the bitwise OR output with a constant 0x3ff is compared to obtain the highest bit of the normalized coding lower limit lowO. In this embodiment, if the result of the bitwise OR operation is equal to 0x3ff, then the highest bit of the normalized coding lower limit low0 is 1, otherwise it is 0.

After the normalized encoding lower limit lowO passes through the second D flip-flop 29, the output encoding lower limit low is to be updated with the normalization module 20 for standby in the next cycle.

The structure and working flow of the coding lower limit update and normalization module 20 in the conventional coding mode and the cut-off coding mode of the embodiment of the present invention will be described in detail below with reference to FIG. 7 . As shown in FIG. 7 , the encoding lower limit update and normalization module 20 in the normal encoding mode and the cut-off encoding mode includes a third selector 21, a first adder 23, a fifth selector 24, and a second shifter. 25 , a third shifter 26 , a fourth shifter 27 , a first comparator 28 and a second D flip-flop 29 .

Wherein, the input of the third selector 21 is the interval length rMPS of the high-probability character outputted by the encoding interval update and normalization module 10 and a default value of an encoding lower limit (which can be 0 here), and the selection factor is an encoding symbol bin, The output is low_add1. Specifically, when the encoding symbol bin is a low-probability character rLPS, the output low_add1 of the third selector 21 is equal to rMPS; otherwise, the output low_add1 is equal to 0.

Wherein, the first adder 23 obtains the encoding lower limit low _nor before normalization according to the third selector 21 and the encoding lower limit low.

Wherein, the fifth selector 24 obtains the iteration number S1 of encoding lower limit normalization according to the encoding mode mode. The input of the fifth selector 24 is the default value of the number of iterations S of the encoding interval normalization and the number of iterations of the encoding lower limit normalization, which can be 1 here, the selection factor is the encoding mode mode, and the output is the encoding lower limit normalization The number of iterations S1. The number of iterations S1 of encoding lower bound normalization is equal to S.

Wherein, the second shifter 25 shifts the coding lower limit low _nor before normalization to the left by S1 bits, and takes the lower 9 bits to obtain the lower 9 bits of the normalized coding lower limit low0, and the second shifter 25 The obtained result is shifted right by 9 bits by the third shifter 26 to obtain the overflow bit overflow_low of the encoding lower limit. The overflow bit overflow_low of the encoding lower limit is input to the next-level pipeline engine, that is, the code stream extraction module 30 . In this embodiment, the number of overflow bits overflow_low of the encoding lower limit is 8 bits, but it should be understood that the present invention does not limit the number of overflow bits overflow_low of the encoding lower limit.

The highest bit of the normalized coding lower limit lowO is obtained by the following method: judge whether the 10th bit of low _nor and the subsequent S1 bits are all 1, and if they are all 1, then the normalized coding lower limit The highest bit of lowO is 1, otherwise it is 0. The specific implementation method is as follows, the fourth shifter 27 shifts the constant 0x1ff to the left by S1 bits and outputs the bitwise OR output with the lower limit low _nor of the encoding before normalization, and the first comparator 28 combines the result of the bitwise OR output with a constant 0x3ff Compare to obtain the highest bit of the normalized coding lower limit lowO. In this embodiment, if the result of the bitwise OR operation is equal to 0x3ff, then the highest bit of the normalized coding lower limit low0 is 1, otherwise it is 0.

After the normalized encoding lower limit lowO passes through the second D flip-flop 29, the output encoding lower limit low is to be updated with the normalization module 20 for standby in the next cycle.

The structure and working process of the coding lower limit update and normalization module 20 in the equal-probability coding mode of the embodiment of the present invention will be described in detail below with reference to FIG. 8 . As shown in FIG. 8 , the coding lower limit update and normalization module 20 in the equal probability coding mode includes a third selector 21, a first adder 23, a fifth selector 24, a second shifter 25, a Three shifters 26 , a fourth shifter 27 , a first comparator 28 and a second D flip-flop 29 .

Wherein, the input of the third selector 21 is the current encoding interval range output by the encoding interval update and normalization module 10 and the default value of the encoding lower limit (here may be 0), the selection factor is the encoding symbol bin, and the output is low_add1 . Specifically, when the encoding symbol bin is 1, the output low_add1 is equal to range; otherwise, the output low_add1 is equal to 0.

Wherein, the first adder 23 obtains the coding lower limit low _nor before normalization according to the third selector 21 and twice the coding lower limit (ie, 2*low).

Wherein, the fifth selector 24 obtains the iteration number S1 of encoding lower limit normalization according to the encoding mode mode. The input of the fifth selector 24 is the default value of the number of iterations S of the encoding interval normalization and the number of iterations of the encoding lower limit normalization, which can be 1 here, the selection factor is the encoding mode mode, and the output is the encoding lower limit normalization The number of iterations S1 of the encoding lower limit normalization is equal to 1.

Wherein, the second shifter 25 shifts the coding lower limit low _nor before normalization to the left by S1 bits, and takes the lower 9 bits to obtain the lower 9 bits of the normalized coding lower limit low0, and the second shifter 25 The obtained result is shifted right by 9 bits by the third shifter 26 to obtain the overflow bit overflow_low of the encoding lower limit. The overflow bit overflow_low of the encoding lower limit is input to the next-level pipeline engine, that is, the code stream extraction module 30 . In this embodiment, the number of overflow bits overflow_low of the encoding lower limit is 8 bits, but it should be understood that the present invention does not limit the number of overflow bits overflow_low of the encoding lower limit.

Among them, the highest bit of the normalized coding lower limit lowO is obtained by the following method: judging whether the 10th bit of the lownor and the subsequent S1 bits are all 1, and if they are all 1, the normalized coding The highest bit of the lower bound lowO is 1, otherwise it is 0. The specific implementation is as follows, the fourth shifter 27 shifts the constant 0x1ff to the left by S1 bits and outputs the bit-wise OR output with the lower limit lownor before normalization, and the first comparator 28 performs bit-wise OR output with a constant 0x3ff Compare to obtain the highest bit of the normalized encoding lower limit lowO. In this embodiment, if the result of the bitwise OR operation is equal to 0x3ff, then the highest bit of the normalized coding lower limit low0 is 1, otherwise it is 0.

After the normalized encoding lower limit lowO passes through the second D flip-flop 29, the output encoding lower limit low is to be updated with the normalization module 20 for standby in the next cycle.

Please refer to FIG. 9 , which is a schematic structural diagram of a code stream extraction module according to an embodiment of the present invention. As shown in Figure 9, the code stream extraction module 30 includes the first 0 detection circuit 31, the sixth selector 32, the third D flip-flop 33, the seventh selector 34 and a code stream output unit, the code stream output unit An eighth selector 35 , a fifth shifter 36 and a sixth shifter 37 are included. Among them, the first 0 detection circuit 31, the sixth selector 32, the third D flip-flop 33 and the seventh selector 34 are used to output the code stream length bit_len, the eighth selector 35, the fifth shifter 36 and the sixth shifter The bit device 37 is used to output the code stream bit_stream.

The first 0 detection circuit 31 detects the first 0 at the end of the overflow bit overflow_low of the encoding lower limit output by the upper-level pipeline engine, and outputs the number N of 1s that appear after the first 0, and the number of 1s that appear after the first 0. The number N is compared with the iteration number S1 of the encoding lower limit normalization, and the comparison result sel is obtained. Here, the comparison result sel indicates the size relationship between S1 and N in code stream extraction. When S1 is smaller than N, sel is 0; when S1 is equal to N, sel is 1; when S1 is greater than N, sel is 2.

The sixth selector 32 selects according to the comparison result sel to obtain the number of uncertain bits num_ostd. The input of the sixth selector 32 is N, and the default value of the number of bits is uncertain, which can be 0 and S1+num_ostd here, and the selection factor is the comparison result sel. When the comparison result sel is 0, the number of uncertain bits num_ostd obtained by the sixth selector 32 is S1+num_ostd; when the comparison result sel is 1, the number of uncertain bits num_ostd obtained is 0; when the comparison result sel is 2, the obtained number of uncertain bits num_ostd is N.

The number of uncertain bits num_ostd output by the sixth selector 32 is input to the D terminal of the third D flip-flop 33, and the third D flip-flop 33 stores the number of uncertain bits num_ostd as a backup for the next cycle, and the third D flip-flop The output of the Q terminal of the device 33 is the number of uncertain bits num_ostd in the current period. The uncertain bit refers to BitOutStanding in the H.264 reference code, which is counted by num_ostd when it is not possible to determine whether the final output is '1' or '0'. The number of uncertain bits num_ostd is used to output the code stream length bit_len of the current period and calculate the new number of uncertain bits num_ostd.

The seventh selector 34 outputs the code stream length bit_len according to the comparison result sel. The input of the seventh selector 34 is S1+num_ostd, the default value of the code stream length (here may be 0), S1+num_ostd-N, and the selection factor is the comparison result sel. When the comparison result sel is 0, the output code stream length bit_len is S1+num_ostd; when the comparison result sel is 1, the output code stream length bit_len is 0; when the comparison result sel is 2, the output code stream length bit_len It is S1+num_ostd-N.

It can be seen that when sel is 0, the code stream length bit_len is updated to S1+num_ostd, and the number of uncertain bits num_ostd is updated to 0; when sel is 1, the code stream length bit_len is updated to 0, and the number of uncertain bits num_ostd is updated to S1+num_ostd; when sel is 2, it means that S1 is greater than N, the code stream length bit_len is updated to S1+num_ostd–N, and the number of uncertain bits num_ostd is updated to N.

In the code stream output unit, the input of the eighth selector 35 is two constants, which are 64'h7fffffffffffffff and 64'h8000000000000000 respectively, and the selection factor of the eighth selector 35 is the highest bit overflow_low[S1] of the effective code stream. When the highest bit overflow_low[S1] of the effective code stream is 0, the output is 64’h7fffffffffffffff; when the highest bit overflow_low[S1] of the effective code stream is 1, the output is 64’h8000000000000000. The output result of the eighth selector 35 is shifted right by 63-num_ostd bits through the fifth shifter 36, then shifted left by S1+1 bits through the sixth shifter 37, and finally with the last bit overflow_low[s1-1:0 ] bitwise OR output.

As a result, the code stream output unit can output the highest bit overflow_low[S1] of the code stream and the middle bit of the uncertain number of bits num_ostd in sequence according to the overflow bit overflow_low of the lower limit of the encoding input by the upper-level pipeline engine in a left-aligned manner! overflow_low[S1] and several last overflow_low[s1-1:0]. Among them, overflow_low[S1] represents the S1+1th bit of the overflow bit overflow_low of the encoding lower limit, ! overflow_low[S1] indicates the inversion of the S1+1 bit of the overflow bit overflow_low of the encoding lower limit, and overflow_low[s1-1:0] indicates the lower S1 bit of the overflow bit overflow_low of the encoding lower limit. The code stream bit_stream output by the code stream output unit has accommodated various situations and conforms to the H.264/AVC standard. The code stream extraction module 20 can complete the code stream extraction operation within one clock cycle, the code stream extraction operation is very simple, and no additional FIFO resource (data cache) is needed.

As can be seen from the above, in the binary arithmetic encoder provided by the embodiment of the present invention, the encoding interval update and normalization module 10 and the encoding lower limit update and normalization module 20 realize the sharing of the three encoding modes through the selector without adding any The control circuit can reduce hardware cost. Moreover, since the number of iterations S1 normalized by the encoding lower limit and the overflow bit overflow_low of the encoding lower limit can be normalized by one or more encoding symbols bin binary arithmetic encoding, the binary arithmetic encoder can The code stream extraction operation of one or more encoding symbol bins is completed, and the code stream extraction operation is very simple, without additional FIFO resources (data cache).

Correspondingly, this embodiment also provides a coding method of a binary arithmetic coder.

Please refer to FIG. 10 , which is a flowchart of an encoding method of a binary arithmetic encoder according to an embodiment of the present invention. As shown in Figure 10, and in conjunction with Figure 1, the coding method of described binary arithmetic coder comprises the following steps:

Step S10: The encoding interval update and normalization module 10 acts as the first-level pipeline engine to extract the encoding state, encoding mode and encoding symbols, update the encoding interval and complete the normalization of the encoding interval, and output the iteration of the normalization of the encoding interval The number of times, the encoding interval, and the interval length of high-probability characters;

Step S11: The encoding lower limit update and normalization module 20 is used as the second-level pipeline engine to extract encoding symbols, encoding modes, the number of iterations of encoding interval normalization, encoding intervals, and the interval length of high-probability characters, and update the encoding lower limit and complete Normalize the lower limit of encoding, and output the overflow bits of the lower limit of encoding and the number of iterations of normalization of the lower limit of encoding;

Step S12: The code stream extraction module 30, as a third-stage pipeline engine, extracts the number of iterations normalized by the encoding lower limit and overflow bits of the encoding lower limit, and outputs a variable-length code stream and the length of the code stream.

Specifically, please refer to FIG. 11. In this embodiment, step S10 further includes:

Step S101: the first selector 11 obtains the interval length rLPS of the low-probability character in the current encoding mode according to the encoding mode;

Specifically, the coding mode mode includes a normal coding mode, an equal probability coding mode, and a cutoff coding mode; encoding mode. When the encoding mode mode is 0 or 1, that is, in the normal encoding mode or the equal probability encoding mode, the interval length rLPS of the low-probability character is obtained according to the result of the probability state state table lookup and range[7:6]. The probability state state looks up the table to obtain 4 rLPS initial values (rLPS0, rLPS1, rLPS2, rLPS3), and the first selector 11 selects one of them as the interval length rLPS of the low-probability character according to range[7:6] as output. When the encoding mode mode is 2, that is, in the cut-off encoding mode, the interval length rLPS of the output low-probability character is equal to 2;

Step S102: The first subtractor 12 obtains the interval length rMPS of the high-probability character in the current encoding mode according to the current encoding interval range and the interval length rLPS of the low-probability character in the current encoding mode; specifically, the rMPS calculation formula is: rMPS = range-rLPS;

Step S103: The second selector 12 selects from the current coding interval range, the interval length rLPS of low-probability characters in the current coding mode, or the interval rMPS of high-probability characters in the current coding mode according to the current coding mode mode and the current coding symbol bin Choose one of the lengths as the encoding interval range _nor before normalization to output; in this embodiment, if the current encoding mode mode is 1, select the current encoding interval range as output; if the current encoding mode mode is 2 and the current encoding symbol When the bin is 1, or the current encoding mode mode is 0 and the current encoding symbol bin is 0, select the interval length rLPS of the low-probability character in the current encoding mode as the output; in other cases except the above two, select the current encoding mode The interval length rMPS of the following high probability characters is output;

Step S104: The first 1 detection circuit 15 counts the leading 0s of the encoding interval range _nor before normalization to obtain the iteration number S of the encoding interval normalization; the iteration number S of the encoding interval normalization will be input to the next Level pipeline engine, that is, the encoding lower limit update and normalization module 20;

Step S105: The first shifter 16 performs left shift on the coding interval range _nor before normalization according to the iteration number S of the normalization of the coding interval and outputs the normalized coding interval rangeO, and rangeO passes through the first D After flip-flop 17, the output encoding interval range is reserved for the next cycle.

Specifically, please refer to FIG. 12. In this embodiment, step S11 further includes:

Step S111: the third selector 21 selects from the current encoding range range, the interval length rMPS of the high probability character in the current encoding mode or the default value of the encoding lower limit according to the current encoding mode mode and the current encoding symbol bin (here can be is 0) to select one as output; in this embodiment, when the encoding mode mode is 0 and the encoding symbol bin is a small probability character rLPS, the low_add1 output by the third selector 21 is equal to rMPS; when the mode is 1 and the bin is 1 When , the output low_add1 is equal to range; otherwise, the output low_add1 is equal to 0;

Step S112: According to the current encoding mode, the fourth selector 22 selects one of the encoding lower limit low or twice the encoding lower limit 2*low as an output; in this embodiment, when the mode is 1, the output of the fourth selector 22 low_add0 is 2*low, select low as output in other cases;

Step S113: the first adder 23 obtains the coding lower limit low _nor before normalization according to the third selector 21 and the fourth selector 22;

Step S114: According to the current encoding mode, the fifth selector 24 selects one from the default value (here, it may be 1) of the number of iterations S normalized by the encoding interval and the number of iterations S1 normalized by the encoding lower limit. It is output as the number of iterations of encoding lower limit normalization; specifically, when mode is 1, that is, in the equal probability encoding mode, the number of iterations S1 of encoding lower limit normalization output by the fifth selector is equal to 1; in other modes , the number of iterations S1 of coding lower limit normalization is equal to S;

Step S115: The second shifter 25 shifts the lower encoding limit before normalization to the left according to the iteration number of the lower encoding limit normalization, and takes the lower 9 bits to obtain the lower 9 bits of the lower encoding limit after normalization;

Step S116: The third shifter 26 shifts the result obtained by the second shifter 25 to the right by 9 bits to obtain the overflow bit overflow_low of the encoding lower limit; the overflow bit overflow_low of the encoding lower limit is input to the next-level pipeline engine, that is, the code stream Extraction module; In the present embodiment, the overflow_low of the encoding lower limit has 8 digits;

Step S117: The fourth shifter 27 shifts the constant 0x1ff to the right according to the number of iterations of the lower limit of encoding normalization, and the result of the bitwise OR operation of the constant _0x1ff and the lower limit of encoding before normalization passes through the first comparator 28 is compared with the constant 0x3ff, and the highest bit of the normalized coding lower limit low0 is obtained according to the comparison result; in this embodiment, if the result of the bitwise OR operation is equal to 0x3ff, then the highest bit of the normalized coding lower limit low0 It is 1, otherwise it is 0; after the normalized encoding lower limit low0 passes through the second D flip-flop 29, the output low is to be used in the next cycle.

Specifically, please refer to FIG. 13, the code stream extraction method S12 includes:

Step 1: Perform trailing zero detection on the overflow bit overflow_low of the encoding lower limit to obtain the number N of 1s that appear after the first 0;

Step 2: Compare the number of iterations S1 of encoding lower limit normalization with the number N of 1s that appear after the first 0, and obtain the comparison result sel;

Step 3: Obtain the number of uncertain bits num_ostd and the length of the output code stream bit_len according to the comparison result sel of S1 and N;

Step 4: Output the variable-length code stream.

Wherein, if S1 is less than N in step 2, it means that the previously accumulated indeterminate bits can be output, and all the current S1 are definite bits and can be output, and the accumulated indeterminate bits are recorded as num_ostd. In step 3, the currently outputtable bit_len is obtained as S1+num_ostd, and the new uncertain number of bits num_ostd is updated to 0. Then output the code stream, first output the highest bit overflow_low[S1], and then output the middle bit of num_ostd bits! overflow_low[S1], and finally output a number of last bits overflow_low[s1-1:0]. So far, the extraction of the current code stream is completed.

If S1 is equal to N in step 2, it means that all current S1 bits are indeterminate bits, therefore, there is currently no outputable bit, and the new indeterminate number of bits num_ostd obtained in step 3 is updated to S1+num_ostd. So far, the extraction of the current code stream is completed.

If S1 is greater than N, it means that there are currently certain bits to output the previous uncertain bits, and there are also new uncertain bits. Therefore, the current output bit length obtained in step 3 is S1+num_ostd–N, and the new number of uncertain bits num_ostd is updated to N. Then output the code stream, first output the highest bit overflow_low[S1], and then output the middle bit of num_ostd bits! overflow_low[S1], and finally output a number of last bits overflow_low[s1-1:0]. So far, the extraction of the current code stream is completed.

It can be seen that the code stream bit_stream consists of three parts, namely the highest bit overflow_low[S1] and the middle bit! overflow_low[S1] and the last bit overflow_low[s1-1:0], the highest bit overflow_low[S1] and the middle bit when the code stream is output! overflow_low[S1] and the last overflow_low[S1-1:0] are output in sequence. When the highest bit is 1, first output 1, followed by 0 of the num_ostd bit, and finally output overflow_low[s1-1:0]. When the highest bit is 0, first output 0, followed by num_ostd bit 1, and finally output overflow_low[s1-1:0].

In the code stream extraction method, the code stream extraction operation of one or more coded symbol bins can be completed in one clock cycle, so no additional FIFO resource (data buffer) is needed.

In the coding method of the binary arithmetic coder, different parameters can be selected according to different coding modes, so as to realize the sharing of various coding modes, including normal coding mode, equal probability coding mode and cut-off coding mode. Moreover, in the code stream extraction method, both the normalized number of iterations S1 and the overflow_low of the encoding lower limit can be obtained after binary arithmetic coding normalization by one or more code symbols bin, therefore, the code stream extraction method The code stream extraction operation of one or more coding symbol bins can be completed in one cycle, and the processing speed of binary arithmetic coding can be effectively improved.

To sum up, in the binary arithmetic encoder and its encoding method provided by the embodiment of the present invention, the encoding interval update and normalization module, the encoding lower limit update and normalization module respectively use selectors to adapt to various encoding modes, The binary arithmetic coder can be efficiently multiplexed and the hardware cost is reduced. At the same time, the code stream extraction method can complete the code stream extraction of one or more bits in one cycle, and the code stream extraction operation is simple without additional FIFO resources (data cache).

It should be noted that each embodiment in this specification is described in a progressive manner, each embodiment focuses on the differences from other embodiments, and the same and similar parts of each embodiment can be referred to each other. As for the method disclosed in the embodiment, since it corresponds to the system disclosed in the embodiment, the description is relatively simple, and for the relevant part, please refer to the description of the system part.

The above description is only a description of the preferred embodiments of the present invention, and does not limit the scope of the present invention. Any changes and modifications made by those of ordinary skill in the field of the present invention based on the above disclosures shall fall within the protection scope of the claims.

Claims (57)

1. A binary arithmetic coder is used for outputting binary arithmetic coded code streams of multiple coding modes and is characterized by comprising a coding interval updating and normalizing module, a coding lower limit updating and normalizing module and a code stream extracting module, wherein the coding interval updating and normalizing module, the coding lower limit updating and normalizing module and the code stream extracting module form a parallel three-level flow structure;

the coding interval updating and normalizing module is used as a first-stage pipeline engine and is used for extracting a coding state, a coding mode and a coding symbol, updating a coding interval, completing the normalization of the coding interval and outputting the iteration times of the normalization of the coding interval and the coding interval to the coding lower limit updating and normalizing module;

the code lower limit updating and normalizing module is used as a second-stage pipeline engine and is used for extracting a code symbol, a code mode, the iteration times of code interval normalization and a code interval, updating the code lower limit, completing the normalization of the code lower limit, and outputting the overflow bit of the code lower limit and the iteration times of the code lower limit normalization to the code stream extracting module;

and the code stream extraction module is used as a third-level pipeline engine and is used for extracting the iteration times of the normalization of the coding lower limit and the overflow bit of the coding lower limit and outputting the variable length code stream and the code stream length.

2. The binary arithmetic encoder of claim 1, wherein the encoding interval update and normalization module further outputs an interval length of a high probability character to the encoding lower limit update and normalization module.

3. The binary arithmetic encoder of claim 2, wherein the binary arithmetic encoder implements mode multiplexing through a selector.

4. The binary arithmetic encoder of claim 3, wherein the coding modes include a regular coding mode, an equiprobable coding mode, and a cutoff coding mode.

5. The binary arithmetic coder according to claim 4, wherein the coding section updating and normalizing module comprises a first selector, a first subtractor, a second selector, a first 1 detection circuit, a first shifter and a first D flip-flop when the mode is multiplexed; wherein,

the first selector is used for obtaining the interval length of the small probability character in the current coding mode according to the extracted coding mode;

the first subtracter is used for obtaining the interval length of the large probability character in the current coding mode according to the current coding interval and the interval length of the small probability character in the current coding mode;

the second selector is used for selecting one of the current coding interval, the interval length of the small probability character in the current coding mode and the interval length of the large probability character in the current coding mode as a coding interval before normalization according to the current coding mode and the current coding symbol and outputting the coding interval;

a first 1 detection circuit, configured to count leading 0 of the coding interval before normalization to obtain the number of iterations of coding interval normalization;

and the first shifter is used for performing left shift on the coding interval before normalization according to the iteration times of the coding interval normalization and outputting the normalized coding interval, and the normalized coding interval passes through the first D trigger and then outputs the coding interval to be reserved in the next period.

6. The binary arithmetic encoder as claimed in claim 5, wherein when the current encoding mode is the regular encoding mode or the equal probability encoding mode, table lookup is performed to obtain initial interval lengths of 4 small probability characters, and the first selector selects one of the initial interval lengths of 4 small probability characters as the interval length of the small probability character in the current encoding mode; and when the current coding mode is the cut-off coding mode, the interval length of the small probability character is equal to 2.

7. The binary arithmetic encoder of claim 5, wherein the second selector selects a current coding section as an output when the current coding mode is an equiprobable coding mode; when the current coding mode is a cut-off coding mode and the current coding symbol is 1, or the current coding mode is a conventional coding mode and the current coding symbol is 0, the second selector selects the interval length of the small probability character in the current coding mode as output; otherwise, the second selector selects the interval length of the high-probability character in the current coding mode as output.

8. The binary arithmetic encoder of claim 2, wherein when the encoding mode is a normal encoding mode, the encoding section updating and normalizing module includes a first subtractor, a second selector, a first 1 detection circuit, a first shifter, and a first D flip-flop; wherein,

the first subtracter is used for obtaining the interval length of the large probability character in the current coding mode according to the current coding interval and the interval length of the small probability character in the current coding mode;

the second selector is used for selecting one of the interval length of the small probability character in the current coding mode or the interval length of the large probability character in the current coding mode as a coding interval before normalization to output according to the current coding mode and the current coding symbol;

a first 1 detection circuit, configured to count leading 0 of the coding interval before normalization to obtain the number of iterations of coding interval normalization;

and the first shifter is used for performing left shift on the coding interval before normalization according to the iteration times of the coding interval normalization and outputting the normalized coding interval, and the normalized coding interval passes through the first D trigger and then outputs the coding interval to be reserved in the next period.

9. The binary arithmetic coder of claim 2, wherein when the coding mode is an equiprobable coding mode, the coding section updating and normalizing module comprises a first trigger, the coding section is directly output as a normalized coding section, and the normalized coding section is output as a coding section for the next cycle after passing through the first D trigger.

10. The binary arithmetic encoder of claim 2, wherein when the encoding mode is a cut-off encoding mode, the encoding section updating and normalizing module includes a first subtractor, a second selector, a first 1 detection circuit, a first shifter, and a first D flip-flop; wherein,

the first subtracter is used for obtaining the interval length of the large probability character in the current coding mode according to the current coding interval and the interval length of the small probability character in the current coding mode;

the second selector is used for selecting one of the interval length of the small probability character in the current coding mode or the interval length of the large probability character in the current coding mode as a coding interval before normalization to output according to the current coding mode and the current coding symbol;

a first 1 detection circuit, configured to count leading 0 of the coding interval before normalization to obtain the number of iterations of coding interval normalization;

and the first shifter is used for performing left shift on the coding interval before normalization according to the iteration times of the coding interval normalization and outputting the normalized coding interval, and the normalized coding interval passes through the first D trigger and then outputs the coding interval to be reserved in the next period.

11. The binary arithmetic encoder of claim 4 wherein the code lower limit update and normalization module includes a third selector, a fourth selector, a first adder, a fifth selector, a second shifter, a third shifter, a fourth shifter, a first comparator, and a second D flip-flop when the mode is multiplexed; wherein,

a third selector, configured to select one of the current coding interval, an interval length of a high-probability character in the current coding mode, and a default value of a lower coding limit as an output according to the current coding mode and the current coding symbol;

a fourth selector, for selecting one of the lower coding limit or 2 times lower coding limit as output according to the current coding mode;

the first adder is used for obtaining a coding lower limit before normalization according to the third selector and the fourth selector;

a fifth selector, configured to select one of the default values of the number of iterations of the coding interval normalization and the number of iterations of the coding lower limit normalization according to the coding mode, and output the selected one as the number of iterations of the coding lower limit normalization;

the second shifter is used for shifting the lower coding limit before normalization to the left according to the iteration number of the lower coding limit normalization, and taking the lower 9 bits to obtain the lower 9 bits of the lower coding limit after normalization;

a third shifter, configured to shift the result obtained by the second shifter by 9 bits to the right to obtain an overflow bit of the lower coding limit;

and the fourth shifter is used for shifting the constant 0x1ff to the right according to the iteration number of the normalization of the lower coding limit, comparing the result of bit-or-operation between the right-shifted constant 0x1ff and the lower coding limit before the normalization with the constant 0x3ff through the first comparator, obtaining the highest bit of the lower coding limit after the normalization according to the comparison result, and outputting the lower coding limit to be reserved in the next period after the lower coding limit after the normalization passes through the second D trigger.

12. The binary arithmetic encoder of claim 11, wherein the third selector selects the interval length of the high probability character as the output when the current coding mode is the normal coding mode and the coded symbol is the interval length of the low probability character; when the current coding mode is the equal probability coding mode and the coding symbol is 1, the third selector selects the current coding interval as output; and in other cases, the third selector selects the default value of the lower coding limit as output.

13. The binary arithmetic encoder of claim 11, wherein the fourth selector selects a lower coding limit of 2 times as output when the current coding mode is the equal probability coding mode, and otherwise selects the lower coding limit as output.

14. The binary arithmetic encoder of claim 11, wherein the fifth selector selects a default value of the number of iterations of the encoding lower limit normalization as an output when the current encoding mode is the equal probability encoding mode, and otherwise the fifth selector selects the number of iterations of the encoding section normalization as an output.

15. The binary arithmetic encoder of claim 11, wherein the fourth shifter shifts a constant 0x1ff right by a bit of a lower code limit normalized iteration number and outputs the bit or the output with a lower code limit before the normalization, the first comparator compares the result of the bit or the output with a constant 0x3ff, and if the result of the bit or the operation is equal to 0x3ff, the highest bit of the lower code limit after the normalization is 1, otherwise is 0.

16. The binary arithmetic encoder of claim 2, wherein when the encoding mode is a regular encoding mode and/or a cut-off encoding mode, the encoding lower limit update and normalization module comprises: the first selector, the second adder, the third selector, the second shifter, the third shifter, the fourth shifter, the first comparator and the second D trigger; wherein,

the third selector is used for selecting the interval length of the large-probability character in the current coding mode and the default value of the lower coding limit to output according to the current coding symbol;

the first adder is used for obtaining a coding lower limit before normalization according to the coding lower limit of the third selector;

a fifth selector, configured to select, according to the coding mode, the iteration number normalized by the coding interval as the iteration number normalized by the coding lower limit, and output the iteration number;

the second shifter is used for shifting the lower coding limit before normalization to the left according to the iteration number of the lower coding limit normalization, and taking the lower 9 bits to obtain the lower 9 bits of the lower coding limit after normalization;

a third shifter, configured to shift the result obtained by the second shifter by 9 bits to the right to obtain an overflow bit of the lower coding limit;

and the fourth shifter is used for shifting the constant 0x1ff to the right according to the iteration number of the normalization of the lower coding limit, comparing the result of bit-or-operation between the right-shifted constant 0x1ff and the lower coding limit before the normalization with the constant 0x3ff through the first comparator, obtaining the highest bit of the lower coding limit after the normalization according to the comparison result, and outputting the lower coding limit to be reserved in the next period after the lower coding limit after the normalization passes through the second trigger.

17. The binary arithmetic encoder of claim 2, wherein when the encoding mode is an equiprobable encoding mode, the encoding lower limit updating and normalizing module comprises a third selector, a first adder, a fifth selector, a second shifter, a third shifter, a fourth shifter, a first comparator, and a second D flip-flop; wherein,

a third selector, configured to select, according to a current coding symbol, a default value of the current coding interval and a default value of a coding lower limit as an output;

the first adder is used for obtaining a coding lower limit before normalization according to the third selector and the 2 times coding lower limit;

the fifth selector is used for selecting a default value of the iteration times of the coding lower limit normalization as the iteration times of the coding lower limit normalization to output according to the coding mode;

the second shifter is used for shifting the lower coding limit before normalization to the left according to the iteration number of the lower coding limit normalization, and taking the lower 9 bits to obtain the lower 9 bits of the lower coding limit after normalization;

a third shifter, configured to shift the result obtained by the second shifter by 9 bits to the right to obtain an overflow bit of the lower coding limit;

and the fourth shifter is used for shifting the constant 0x1ff to the right according to the iteration number of the normalization of the lower coding limit, comparing the result of bit-or-operation between the right-shifted constant 0x1ff and the lower coding limit before the normalization with the constant 0x3ff through the first comparator, obtaining the highest bit of the lower coding limit after the normalization according to the comparison result, and outputting the lower coding limit to be reserved in the next period after the lower coding limit after the normalization passes through the second D trigger.

18. The binary arithmetic encoder according to claim 1, wherein the code stream extraction module includes a first 0 detection circuit, a sixth selector, a third D flip-flop, a seventh selector, and a code stream output unit; wherein,

the first 0 detection circuit is used for carrying out tail first 0 detection on the overflow bit of the coding lower limit and outputting the number of 1 appearing after the first 0, and the number of 1 appearing after the first 0 is compared with the iteration number of the coding lower limit normalization to obtain a comparison result;

a sixth selector, configured to select one of the number of 1 s appearing after the first 0, a default value of the number of uncertain bits, and a sum of the number of iterations of the encoding lower limit normalization and the number of uncertain bits as the number of uncertain bits according to the comparison result, and output the selected number;

the third D flip-flop is used for storing the number of the uncertain bits output by the sixth selector as a spare of the next period;

a seventh selector, configured to select, according to the comparison result, one of a sum of the number of iterations of the coding lower limit normalization and the number of uncertain bits, a default value of the code stream length, and a difference between the sum of the number of iterations of the coding lower limit normalization and the number of uncertain bits and the number of 1 s appearing after the first 0 as the code stream length to output;

and the code stream output unit is used for outputting the variable length code stream according to the overflow bit of the coding lower limit.

19. The binary arithmetic encoder of claim 18, wherein when the number of iterations of the encoding lower limit normalization is less than the number of 1's occurring after the first 0, the comparison result is 0; when the iteration number of the encoding lower limit normalization is equal to the number of 1 appearing after the first 0, the comparison result is 1; when the number of iterations of the encoding lower limit normalization is greater than the number of 1's appearing after the first 0, the comparison result is 2.

20. The binary arithmetic encoder of claim 19, wherein when the comparison result is 0, the sixth selector outputs a sum of the number of iterations of the encoding lower limit normalization and the number of uncertain bits; when the comparison result is 1, the sixth selector takes the default value of the number of uncertain bits as output; when the comparison result is 2, the sixth selector takes as output the number of 1's appearing after the first 0.

21. The binary arithmetic encoder of claim 20, wherein when the comparison result is 0, the code stream length output by the seventh selector is a sum of the number of iterations of the coding lower limit normalization and the number of uncertain bits; when the comparison result is 1, the code stream length output by the seventh selector is a default value of the code stream length; and when the comparison result is 2, the code stream length output by the seventh selector is the difference between the sum of the iteration times of the coding lower limit normalization and the number of the uncertain bits and the number of 1 s appearing after the first 0.

22. The binary arithmetic encoder of claim 18, wherein the code stream output unit includes an eighth selector, a fifth shifter, and a sixth shifter; when the highest bit of the effective code stream is 0, the eighth selector outputs a first constant, and when the highest bit of the effective code stream is 1, the eighth selector outputs a second constant; and the output result of the eighth selector is shifted to the right by a fifth shifter according to the difference value between 63 and the number of uncertain bits, the output result of the fifth shifter is shifted to the left by the sixth shifter according to the sum of the iteration times of the lower limit normalization of the code and 1, and finally the output result and the last bit are bitwise or output.

23. The binary arithmetic encoder of claim 22, wherein the code stream output unit sequentially outputs a highest bit, a plurality of middle bits, and a plurality of last bits in a left-aligned manner; when the highest bit is 1, the middle bits are all 0, when the highest bit is 0, the middle bits are all 1, and the number of the middle bits is equal to the number of the uncertain bits.

24. The binary arithmetic encoder of claim 23, wherein the stream extraction module performs a stream extraction operation in one clock cycle.

25. A coding interval updating and normalizing module is characterized by being used for extracting a coding state, a coding mode and a coding symbol, updating a coding interval, completing normalization of the coding interval and outputting iteration times of normalization of the coding interval and the coding interval;

the coding interval updating and normalizing module comprises a first selector, a first subtracter, a second selector, a first 1 detection circuit, a first shifter and a first D trigger; wherein,

the first selector is used for obtaining the interval length of the small probability character in the current coding mode according to the extracted coding mode;

the first subtracter is used for obtaining the interval length of the large probability character in the current coding mode according to the current coding interval and the interval length of the small probability character in the current coding mode;

the second selector is used for selecting one of the current coding interval, the interval length of the small probability character in the current coding mode and the interval length of the large probability character in the current coding mode as a coding interval before normalization according to the current coding mode and the current coding symbol and outputting the coding interval;

a first 1 detection circuit, configured to count leading 0 of the coding interval before normalization to obtain the number of iterations of coding interval normalization;

and the first shifter is used for performing left shift on the coding interval before normalization according to the iteration times of the coding interval normalization and outputting the normalized coding interval, and the normalized coding interval passes through the first D trigger and then outputs the coding interval to be reserved in the next period.

26. The encoding section updating and normalizing module of claim 25, wherein when the current encoding mode is the regular encoding mode or the equal probability encoding mode, the initial values of the section lengths of the 4 small probability characters are obtained by table lookup, and the first selector selects one of the initial values of the section lengths of the 4 small probability characters as the section length of the small probability character in the current encoding mode; and when the current coding mode is the cut-off coding mode, the interval length of the small probability character is equal to 2.

27. The encoding section updating and normalizing module of claim 25, wherein the second selector selects the current encoding section as the output when the current encoding mode is the equal probability encoding mode; when the current coding mode is a cut-off coding mode and the current coding symbol is 1, or the current coding mode is a conventional coding mode and the current coding symbol is 0, the second selector selects the interval length of the small probability character in the current coding mode as output; otherwise, the second selector selects the interval length of the high-probability character in the current coding mode as output.

28. A code lower limit updating and normalizing module is characterized by being used for extracting a code symbol, a code mode, the iteration number of code interval normalization and a code interval, updating the code lower limit, completing the normalization of the code lower limit, and outputting an overflow bit of the code lower limit and the iteration number of the code lower limit normalization;

the coding lower limit updating and normalizing module comprises a third selector, a fourth selector, a first adder, a fifth selector, a second shifter, a third shifter, a fourth shifter, a first comparator and a second D trigger; wherein,

a third selector, configured to select one of the current coding interval, an interval length of a high-probability character in the current coding mode, and a default value of a lower coding limit as an output according to the current coding mode and the current coding symbol;

a fourth selector, for selecting one of the lower coding limit or 2 times lower coding limit as output according to the current coding mode;

the first adder is used for obtaining a coding lower limit before normalization according to the third selector and the fourth selector;

a fifth selector, configured to select one of the default values of the number of iterations of the coding interval normalization and the number of iterations of the coding lower limit normalization according to the coding mode, and output the selected one as the number of iterations of the coding lower limit normalization;

the second shifter is used for shifting the lower coding limit before normalization to the left according to the iteration number of the lower coding limit normalization, and taking the lower 9 bits to obtain the lower 9 bits of the lower coding limit after normalization;

a third shifter, configured to shift the result obtained by the second shifter by 9 bits to the right to obtain an overflow bit of the lower coding limit;

and the fourth shifter is used for shifting the constant 0x1ff to the right according to the iteration number of the normalization of the lower coding limit, comparing the result of bit-or-operation between the right-shifted constant 0x1ff and the lower coding limit before the normalization with the constant 0x3ff through the first comparator, obtaining the highest bit of the lower coding limit after the normalization according to the comparison result, and outputting the lower coding limit to be reserved in the next period after the lower coding limit after the normalization passes through the second D trigger.

29. The code lower limit updating and normalizing module of claim 28, wherein when the current coding mode is the normal coding mode and the coding symbol is the interval length of the small probability character, the third selector selects the interval length of the large probability character as the output; when the current coding mode is the equal probability coding mode and the coding symbol is 1, the third selector selects the current coding interval as output; and in other cases, the third selector selects the default value of the lower coding limit as output.

30. The code lower limit updating and normalizing module of claim 28, wherein when the current coding mode is the equal probability coding mode, the fourth selector selects the 2 times code lower limit as the output, and otherwise selects the code lower limit as the output.

31. The code lower limit updating and normalizing module of claim 28, wherein when the current coding mode is an equiprobable coding mode, the fifth selector selects a default value of the number of iterations of the code lower limit normalization as an output, and otherwise the fifth selector selects the number of iterations of the code interval normalization as an output.

32. The code floor update and normalization module of claim 28, wherein the fourth shifter shifts a constant 0x1ff right by a bit of the normalized iteration number of the code floor and outputs the bit-wise or output with the code floor before normalization, the first comparator compares the result of the bit-wise or output with a constant 0x3ff, if the result of the bit-wise or operation is equal to 0x3ff, the highest bit of the normalized code floor is 1, otherwise it is 0.

33. A code stream extraction module is characterized in that the code stream extraction module is used for extracting the iteration number of the normalization of the lower coding limit and the overflow bit of the lower coding limit and outputting a variable length code stream and the code stream length;

the code stream extraction module comprises a first 0 detection circuit, a sixth selector, a third D trigger, a seventh selector and a code stream output unit; wherein,

the first 0 detection circuit is used for carrying out tail first 0 detection on the overflow bit of the coding lower limit and outputting the number of 1 appearing after the first 0, and the number of 1 appearing after the first 0 is compared with the iteration number of the coding lower limit normalization to obtain a comparison result;

a sixth selector, configured to select one of the number of 1 s appearing after the first 0, a default value of the number of uncertain bits, and a sum of the number of iterations of the encoding lower limit normalization and the number of uncertain bits as the number of uncertain bits according to the comparison result, and output the selected number;

the third D flip-flop is used for storing the number of the uncertain bits output by the sixth selector as a spare of the next period;

a seventh selector, configured to select, according to the comparison result, one of a sum of the number of iterations of the coding lower limit normalization and the number of uncertain bits, a default value of the code stream length, and a difference between the sum of the number of iterations of the coding lower limit normalization and the number of uncertain bits and the number of 1 s appearing after the first 0 as the code stream length to output;

and the code stream output unit is used for outputting the variable length code stream according to the overflow bit of the coding lower limit.

34. The code stream extraction module of claim 33, wherein when the number of iterations of the encoding lower limit normalization is less than the number of 1's occurring after the first 0, the comparison result is 0; when the iteration number of the encoding lower limit normalization is equal to the number of 1 appearing after the first 0, the comparison result is 1; when the number of iterations of the encoding lower limit normalization is greater than the number of 1's appearing after the first 0, the comparison result is 2.

35. The code stream extraction module of claim 34, wherein when the comparison result is 0, the sixth selector outputs a sum of the number of iterations of the coding lower limit normalization and the number of uncertain bits; when the comparison result is 1, the sixth selector takes the default value of the number of uncertain bits as output; when the comparison result is 2, the sixth selector takes as output the number of 1's appearing after the first 0.

36. The code stream extraction module of claim 35, wherein when the comparison result is 0, the code stream length output by the seventh selector is a sum of the iteration number of the coding lower limit normalization and the number of uncertain bits; when the comparison result is 1, the code stream length output by the seventh selector is a default value of the code stream length; and when the comparison result is 2, the code stream length output by the seventh selector is the difference between the sum of the iteration times of the coding lower limit normalization and the number of the uncertain bits and the number of 1 s appearing after the first 0.

37. The stream extraction module of claim 36, wherein the stream output unit includes an eighth selector, a fifth shifter, and a sixth shifter; when the highest bit of the effective code stream is 0, the eighth selector outputs a first constant, and when the highest bit of the effective code stream is 1, the eighth selector outputs a second constant; and the output result of the eighth selector is shifted to the right by a fifth shifter according to the difference value between 63 and the number of uncertain bits, the output result of the fifth shifter is shifted to the left by the sixth shifter according to the sum of the iteration times of the lower limit normalization of the code and 1, and finally the output result and the last bit are bitwise or output.

38. The code stream extraction module of claim 37, wherein the code stream output unit sequentially outputs a highest bit, a plurality of middle bits, and a plurality of last bits in a left alignment manner; when the highest bit is 1, the middle bits are all 0, when the highest bit is 0, the middle bits are all 1, and the number of the middle bits is equal to the number of the uncertain bits.

39. The stream extraction module of claim 38, wherein the stream extraction module completes a stream extraction operation in one clock cycle.

40. A coding method of a binary arithmetic coder, the coding method of the binary arithmetic coder comprising:

step S10: the coding interval updating and normalizing module is used as a first-stage pipeline engine, extracts a coding state, a coding mode and a coding symbol, updates a coding interval, completes the normalization of the coding interval, and outputs the iteration times of the normalization of the coding interval, the coding interval and the interval length of a large probability character;

step S11: the coding lower limit updating and normalizing module is used as a second-level pipeline engine, extracts coding symbols, coding modes, the iteration number of coding interval normalization, coding intervals and the interval length of a large probability character, updates the coding lower limit, completes the normalization of the coding lower limit, and outputs overflow bits of the coding lower limit and the iteration number of the coding lower limit normalization;

step S12: and the code stream extraction module is used as a third-level flow engine, extracts the iteration times of the normalization of the coding lower limit and the overflow bit of the coding lower limit, and outputs the variable-length code stream and the code stream length.

41. The encoding method of binary arithmetic encoder as claimed in claim 40, wherein in step S10, said encoding section updating and normalizing module further outputs a section length of a large probability character to said encoding lower limit updating and normalizing module.

42. The encoding method of the binary arithmetic encoder of claim 41, wherein the coding modes include a regular coding mode, an equiprobable coding mode, and a cut-off coding mode.

43. The encoding method of the binary arithmetic encoder of claim 42, wherein the step S10 further comprises:

the first selector obtains the interval length of the small probability character in the current coding mode according to the coding mode;

the first subtracter obtains the interval length of the high probability character in the current coding mode according to the current coding interval and the interval length of the low probability character in the current coding mode;

the second selector selects one of the current coding interval, the interval length of the small probability character in the current coding mode or the interval length of the large probability character in the current coding mode as a coding interval before normalization to output according to the current coding mode and the current coding symbol;

the first 1 detection circuit performs leading 0 counting on the coding interval before normalization to obtain the iteration times of coding interval normalization;

and the first shifter shifts the coding interval before normalization to the left according to the iteration number of the coding interval normalization and outputs the coding interval after normalization, and the coding interval after normalization is output to be reserved in the next period after passing through a D trigger.

44. The encoding method of the binary arithmetic encoder as claimed in claim 43, wherein when the current encoding mode is the regular encoding mode or the equal probability encoding mode, the initial interval length values of the 4 small probability characters are obtained by table look-up, and one of the initial interval length values of the 4 small probability characters is selected by the first selector as the interval length of the small probability character in the current encoding mode; and when the current coding mode is the cut-off coding mode, the interval length of the small probability character is equal to 2.

45. The encoding method of a binary arithmetic encoder according to claim 43, wherein said second selector selects a current coding section as an output when a current coding mode is an equal probability coding mode; when the current coding mode is a cut-off coding mode and the current coding symbol is 1, or the current coding mode is a conventional coding mode and the current coding symbol is 0, the second selector selects the interval length of the small probability character in the current coding mode as output; otherwise, the second selector selects the interval length of the high-probability character in the current coding mode as output.

46. The encoding method of the binary arithmetic encoder of claim 41, wherein the step S11 further comprises:

a third selector selects one of the current coding interval, the interval length of the large-probability character in the current coding mode or the default value of the lower coding limit as output according to the current coding mode and the current coding symbol;

the fourth selector selects one of the lower coding limit or the lower coding limit of 2 times as output according to the current coding mode;

the first adder obtains a coding lower limit before normalization according to the third selector and the fourth selector;

a fifth selector selects one of default values of the iteration times of the coding interval normalization and the iteration times of the coding lower limit normalization as the iteration times of the coding lower limit normalization according to the current coding mode and outputs the selected iteration times of the coding interval normalization and the default values;

the second shifter shifts the lower limit of the code before normalization to the left according to the iteration number of the normalization of the lower limit of the code, and takes the lower 9 bits to obtain the lower 9 bits of the lower limit of the code after normalization;

the third shifter shifts the result obtained by the second shifter to the right by 9 bits to obtain an overflow bit of the lower coding limit;

the fourth shifter shifts the constant 0x1ff to the right according to the iteration number of the normalization of the lower coding limit, the result of bit or operation between the right shifted constant 0x1ff and the lower coding limit before the normalization is compared with the constant 0x3ff through the first comparator, and the highest bit of the lower coding limit after the normalization is obtained according to the comparison result;

and after the normalized lower coding limit passes through a third D trigger, outputting the lower coding limit to be reserved in the next period.

47. The encoding method of the binary arithmetic encoder as claimed in claim 46, wherein when the current encoding mode is the regular encoding mode and the encoding symbol is the interval length of the small probability character, said third selector selects the interval length of the large probability character as the output; when the current coding mode is the equal probability coding mode and the coding symbol is 1, the third selector selects the current coding interval as output; and in other cases, the third selector selects the default value of the lower coding limit as output.

48. The encoding method of the binary arithmetic encoder as claimed in claim 46, wherein when the current encoding mode is the equal probability encoding mode, the fourth selector selects a lower limit of 2 times encoding as the output, and otherwise selects the lower limit of encoding as the output.

49. The encoding method of a binary arithmetic encoder according to claim 46, wherein said fifth selector selects a default value of the number of iterations of the encoding lower limit normalization as an output when the current encoding mode is the equal probability encoding mode, and otherwise said fifth selector selects the number of iterations of the encoding section normalization as an output.

50. The encoding method of the binary arithmetic encoder of claim 46, wherein the fourth shifter shifts a constant 0x1ff right by a bit of a normalized iteration number of the lower encoding limit and outputs the bit or the output of the lower encoding limit before the normalization, the first comparator compares the result of the bit or the output with a constant 0x3ff, and if the result of the bit or the operation is equal to 0x3ff, the highest bit of the lower encoding limit after the normalization is 1, otherwise is 0.

51. The encoding method of the binary arithmetic encoder of claim 41, wherein the step S12 further comprises:

the first 0 detection circuit carries out tail first 0 detection on the overflow bit of the coding lower limit, outputs the number of 1 appearing after the first 0, and compares the number of 1 appearing after the first 0 with the iteration number of the coding lower limit normalization;

a sixth selector selects one of the number of 1 appearing after the first 0, the default value of the number of uncertain bits, the iteration times of the coding lower limit normalization and the sum of the number of uncertain bits as the number of uncertain bits to output according to the comparison result;

a third D flip-flop stores the number of indeterminate bits output by the sixth selector;

a seventh selector selects one of the sum of the iteration times of the coding lower limit normalization and the number of the uncertain bits, a default value of the code stream length, the sum of the iteration times of the coding lower limit normalization and the number of the uncertain bits and the difference value between the number of the uncertain bits and the number of 1 after the first 0 as the code stream length to output according to the comparison result;

and the code stream output unit outputs a variable length code stream according to the overflow bit of the coding lower limit.

52. The encoding method of a binary arithmetic encoder according to claim 51, wherein when the number of iterations of said encoding lower limit normalization is smaller than the number of 1's occurring after said first 0, the comparison result is 0; when the iteration number of the encoding lower limit normalization is equal to the number of 1 appearing after the first 0, the comparison result is 1; when the number of iterations of the encoding lower limit normalization is greater than the number of 1's appearing after the first 0, the comparison result is 2.

53. The encoding method of a binary arithmetic encoder according to claim 52, wherein said sixth selector outputs a sum of the number of iterations of said encoding lower limit normalization and the number of uncertain bits when said comparison result is 0; when the comparison result is 1, the sixth selector takes the default value of the number of uncertain bits as output; when the comparison result is 2, the sixth selector takes as output the number of 1's appearing after the first 0.

54. The encoding method of a binary arithmetic encoder according to claim 53, wherein when said comparison result is 0, a code stream length outputted by said seventh selector is a sum of a number of iterations of said encoding lower limit normalization and a number of uncertain bits; when the comparison result is 1, the code stream length output by the seventh selector is a default value of the code stream length; and when the comparison result is 2, the code stream length output by the seventh selector is the difference between the sum of the iteration times of the coding lower limit normalization and the number of the uncertain bits and the number of 1 s appearing after the first 0.

55. The encoding method of the binary arithmetic encoder of claim 54, wherein the eighth selector outputs the first constant when the most significant bit of the valid code stream is 0, and outputs the second constant when the most significant bit of the valid code stream is 1; and the output result of the eighth selector is shifted to the right by a fifth shifter according to the difference value between 63 and the number of uncertain bits, the output result of the fifth shifter is shifted to the left by the sixth shifter according to the sum of the iteration times of the lower limit normalization of the code and 1, and finally the output result and the last bit are bitwise or output.

56. The encoding method of claim 55, wherein said code stream output unit sequentially outputs a highest bit, a plurality of middle bits and a plurality of last bits in a left alignment manner; when the highest bit is 1, the middle bits are all 0, when the highest bit is 0, the middle bits are all 1, and the number of the middle bits is equal to the number of the uncertain bits.

57. The encoding method of claim 56, wherein said code stream extracting module performs a code stream extracting operation within one clock cycle.