KR0129559B1 - Minimum value sorting apparatus for use in a video encoder - Google Patents

Abstract

The local minimum classifier sorts the matching error occurring in the 1-dimensional matching process in small order when the motion processor uses the local minimum value in the motion processor of the image encoding apparatus, and also sorts the motion vectors for the blocks used for the matching error. In order to generate a motion vector of a block having a plurality of local minimums, two sub-classifiers for alternately processing one-dimensional matching errors, a switch for switching two sub-classifiers, and a controller for controlling these switches and the sub-classifiers are provided. Include.

Description

Minimum sorting device

1 is a flowchart of a method for measuring movement using a local minimum value.

2 is a block diagram of a minimum classifying apparatus according to the present invention.

3 is a detailed block diagram of the first or second dependent classifier shown in FIG.

4 is a detailed block diagram of the comparison unit shown in FIG.

5 is a timing diagram of a minimum classifier.

* Explanation of symbols for main parts of the drawings

10,12 subordinate classifier 14 control unit

16, 18, 62-72, 76: switches 20-24, 32-36, 38-60: resistors

26-30: comparator 74: comparator

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion processor of an image encoding apparatus, and more particularly, to a minimum classifier for selecting a local minimum value when estimating a motion using a local minimum value.

Description of the movement amount measurement using the local minimum is disclosed in Patent No. 92-18391 filed by the Applicant on October 7, 1992, with the apparatus and method for the movement amount measurement using the local minimum value. Omit.

As shown in FIG. 1, the moving amount measurement using the local minimum generates cumulative data in the horizontal and vertical directions of the image blocks to be compared (S2), and generates the generated one-dimensional (1D) cumulative data in the search region. The matching error is calculated by comparing the 1D accumulated data (S4), the predetermined number of blocks are selected in the order of the smallest matching error (S6), and the two-dimensional (2D) pixel unit for the plurality of selected blocks. By comparing the blocks with each other (S8), a final moving amount is determined by selecting positions having similarities between the bands.

SUMMARY OF THE INVENTION An object of the present invention is to provide an apparatus capable of obtaining motion vectors for a plurality of blocks having small horizontal and vertical matching errors between blocks in a movable region of a previous frame and a current block. When the classification apparatus estimates the motion using the local minimum in the motion processor of the image encoding apparatus, the classification apparatus sorts the matching error occurring in the one-dimensional matching process in small order, and also sorts the motion vectors for the blocks used for the matching error together. Generating a motion vector of a block having a plurality of local minimums, comprising two slave classifiers for alternately processing one-dimensional matching errors, a switch for switching two slave classifiers, and a controller for controlling these switches and the slave classifiers. .

Hereinafter, with reference to the accompanying drawings the present invention will be described in detail.

2 is a block diagram of a minimum classifying apparatus according to the present invention.

The minimum classifying apparatus includes a first subordinate classifier 10 of the upper path and a second subordinate classifier 12 of the lower path that sort the matching error and the motion vector in order, and input the matching error and the motion vector as a whole control signal. A first switch 16 connecting to the first or second dependent classifier and a second switch connecting the first or second dependent classifier so as to be opposite to the connection of the first switch 16 so that an aligned motion vector is output; (18) and a control unit (14) for providing a control signal for inputting the entire control signal and for causing the matching error and the motion vector to be aligned or outputting the aligned motion vector in the first and second subordinate classifiers. do.

3 is a detailed block diagram of the first or second dependent classifier shown in FIG.

The dependent classifier is composed of a matching error comparator 19 composed of n stages, an x-axis motion vector storage device 37, and a y-axis motion vector alignment device 49.

Each stage of the matching error comparing device for comparing matching errors includes first registers 20, 22, and 24 for storing matching errors to be compared, and an order for storing the aligned matching errors to be compared. 2 registers 32, 34 and 36 are compared with the output of the first register and the output of the second register, and one of the two output signals is output to the first register of the next stage according to the control signal, and the remaining It comprises a comparator 26, 28, 30 which provides a signal to the second register of the same stage and also provides a switching signal of the x- and y-axis motion vector alignment device.

The motion vector alignment device 37 in the x-axis direction includes third registers 38, 40, and 42 that store motion vectors for blocks that generate a matching error to be compared, and an aligned matching error to be compared. Third and fourth according to the switching control signals of the fourth registers 44, 46, 48, and the comparators 26, 28, and 30 that store the motion vector for the block that has generated? It consists of switches 62, 64 and 66 which output one of the motion vectors of the four registers to the third register of the next stage and store the other one in the fourth register of the same stage.

The moved vector alignment device 49 in the y axis direction is configured similarly to the motion vector alignment device 37 in the x axis direction.

4 is a detailed block diagram of the comparison unit illustrated in FIG. 3.

As shown, the comparator uses the outputs of the first and second registers as inputs 1 and 2, respectively, and compares the magnitudes of the two signals, and inputs 1 according to the control signals OP1 and OP2 and the comparison state. And a switch 76 connecting the input 2 to the output 1 and the output 2.

5 shows a timing diagram of a signal performed in the first stage of the minimum classifier of FIG.

Hereinafter, embodiments of the present invention will be described with reference to FIGS. 2 to 5.

When the horizontal and vertical matching error between the block in the movable region of the previous frame and the current block and the motion vector for each block are input to the minimum classifier, the first switch 16 arranges the input signals according to the entire control signal. The first classifier 10 or the second classifier 12 is connected so that the second switch 18 is connected to the sorted first or second classifier to extract the aligned motion vector.

For example, when the switch 16 is operated such that an input signal is input to the first classifier 10, the first subordinate classifier 10 aligns the input signals in the order of least error. On the other hand, the switch 18 is connected to the second slave classifier 12 so that the aligned motion vector is output.

When the matching error and the motion vector dx, dy are input to the first or second dependent classifier, as shown in FIG. 3, the matching error is aligned through a matching error comparison device composed of n stages. , Motion vectors (dx), (dy) are aligned through n number of X-axis motion vector aligners 37 and Y-axis motion vector aligners 49.

When the dependent classifier sorts the input data in order (OP1 = 0), (OP2 = 1), if any matching error is input to the first register 20, the comparator 26 enters the input matching error and the second. Matching errors aligned in register 32 are input by input 1 and input 2 and compared.

At this time, if the matching error of the input 1 is less than or equal to the matching error of the input 2, the matching error of the input 1 is stored in the second register 32 through the output 2, the data stored in the second register 32 is the second It is input to the first register 22 of the stage. At this time, the logic 1 is output as the switch control signal C1 of the switches 62, 68, and 76.

In contrast, if the data of input 1 is larger than the data of input 2, the matching error of input 1 is stored in second register 32 via output 1, and the data stored in second register 32 is maintained as it is. At this time, the switch control signals c1 of the switches 62, 68, and 76 output logic 0 to control switching.

When the horizontal axis motion vector Vy of the block used for the matching error input to the first register 20 is input to the third register 38, the input motion vector and the aligned motion vector stored in the fourth register 44 are input. Is rearranged by the switch 62. That is, when the switching control signal c1 is 1, the motion vector of the third register 38 is input to the fourth register 44, and the motion vector of the fourth register 44 is the third of the second stage. It is input to the register 40. When the switch control signal c1 is 0, the motion vector of the third register 38 of the first stage is input to the third register 40 of the second stage, and the motion vector of the fourth register 44 is It stays the same.

Similarly, when the vertical axis motion vector Vx of the block used for the matching error input to the first register 20 is input to the fifth register 50, the input motion vector and the aligned motion stored in the sixth register 56 are aligned. The motion vector is rearranged by the switch 68. That is, when the switch control signal c1 is 1, the motion vector of the fifth register 50 is input to the sixth register 56, and the motion vector of the sixth register 56 is made of the second stage. 5 is input to the register 52. When the switch control signal c1 is 0, the motion vector of the fifth register 50 of the first stage is input to the fifth register 52 of the second stage, and the motion vector of the sixth register 56 is It stays the same.

As shown in FIG. 5, the timing diagram of the signal flow performed in the first stage is as follows. When the match error path is input, the signal is latched at the rising edge of the clock signal CLK and stored in the first register 20. The value is compared with the value of the second register 32 to generate the switch control signal c1. This controls the switch to provide the signal compared to the next stage.

By the above-described method, the input matching error and the motion vector are arranged in the order of the smallest matching error. After the matching of the matching error and the motion vectors inputted to the subordinate classifier by the predetermined processing unit is completed, the sorted motion vectors are discharged in order by the switching control signal. Table 1 shows the switching control signals in the comparator shown in FIG.

By switching in the manner shown in Table 1, an aligned motion vector can be output.

The minimum classification apparatus according to the present invention aligns the matching errors for a number of blocks obtained in one-dimensional block matching in the order of the smallest size, thereby facilitating a motion vector for several blocks required for the next step (two-dimensional block matching). You can get it.

In particular, since the minimum block device proposed in the present invention has a simple hardware configuration, it is possible to design a very large scale integrated circuit (VLSI), and by using two subordinate classifiers, the blocks required for 2D matching are continuously selected without delay. Can improve the performance of the overall motion processor.

Claims (4)

In order to obtain a motion vector for a plurality of blocks having a small matching error by aligning the horizontal and vertical matching error between the blocks in the movable region of the previous frame and the current block and the position information of the blocks, First means (10) on a first path for aligning the matching error and motion vectors in order; second means (12) on a second path for aligning the matching error and motion vectors in order; First switching means (16) for connecting the matching error and the motion vector to the first or second path by a predetermined control signal; Second switching means (18) connected to the first switching means by the control signal in an opposite path and for outputting an aligned motion vector; And a control means for inputting the control signal and providing a control signal in which the matching error and the motion vector are aligned or output the aligned motion vector in the first and second means. 2. The apparatus of claim 1, wherein the first and second means comprise: n match error comparing means for aligning the horizontal and vertical matching errors in a small order; And n motion vector aligning means for aligning the motion vector according to the switching signal of the matching error comparing means. 3. The apparatus of claim 2, wherein the matching error comparing means comprises: first storing means for storing matching errors to be compared; Second storage means for storing an aligned matching error which is a signal to be compared; comparing the output of the first storage means with the output of the second storage means, and storing the output in the first stage according to the control signal. Means for providing to said means or said second storage means-and for providing said switching signal, wherein said motion vector aligning means stores a motion vector for the block that caused said matching error to be compared. Storage means; Fourth storage means for storing a motion vector for a block that has generated an aligned matching error as said comparison signal; And switching means for outputting one of the motion vectors of the third and fourth storage means to the next stage and storing the other in the fourth storage means in accordance with the switching signal. The minimum classifying apparatus according to claim 3, wherein the horizontal motion vector and the vertical motion vector of the motion vector are processed separately.