JPH02183668A - color image encoding device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Objective of the Invention] (Industrial Application Field) The present invention is an image encoding device related to compression encoding of color still images or moving images.

(Prior Art) Video telephones, electronic cameras, etc. are currently being actively researched and developed. When compressing and encoding color still images or moving images in such devices, one screen is divided into R, G, and B signals, or luminance and color difference signals, and compression processing is performed on each signal. R, G, B
The amount of information (details of the picture) contained in each signal or each signal of luminance and two color differences is usually different, and when encoding, the amount of information (encoded bits) is determined according to the amount of information. compression encoding.

On the other hand, when considering human visual characteristics, the above R2G, B
It is known that the ease of detection by the human eye differs for each signal or luminance and two-color difference signal. For example, in the case of a luminance two-color difference signal, the luminance signal is generally easier to detect by the human eye than the color difference signal (
In general, the amount of information is greater in the luminance signal). Therefore,
When encoding, it is necessary to allocate the amount of code according to the ease of detection.

In this way, it is necessary to allocate the number of encoding bits to each signal according to the fineness of the picture or the ease with which it is detected by the human eye, but conventionally, the R2O and B signals are considered to have the same importance and are encoded. A method for allocating amounts (Saito, et al.: ″
Adaptive discrete cosine transform of color images Ga1n/5
hape vector quantization method"I E85-120.19
1985), or a method in which the code amount is predetermined in a ratio of luminance signal: color difference signal = 2:1, or a method in which the code amount is set so that the S/N ratio of the color difference signal is 6 dB lower than that of the brightness signal. - (Ito, et al.: Variable-length code selection type DPCM encoding method for color image signals 2'' 1986 Institute of Electronics, Information and Communication Engineers Autumn National Conference D-70), etc. are used.

However, these methods do not allocate the amount of code based on the amount of information actually present in the screen to be encoded, and furthermore, it cannot be said that they fully take human visual characteristics into consideration. For example, even though the ratio of the amount of information between the luminance signal and the color difference signal varies considerably from picture to picture, if it is defined as 2:1, blurring due to insufficient code amount of the luminance signal and the amount of code of one color difference signal will occur. Color leakage occurs due to shortage. Also, if the S/N ratio of the color difference signal is set uniformly 6 dB lower than the luminance signal, colors that should be present may be omitted, especially when a picture with a large amount of information in the color difference signal is input. There is a problem that the reproduced image appears dull or the entire screen appears faded, leading to deterioration of subjective evaluation.

(Problem to be solved by the invention) In this way, in the past, coding bit allocation did not take into account the amount of information held by each signal in the screen or the impact each signal has on human vision. Therefore, the subjective evaluation of the reproduced image varies greatly depending on the input image, and the quality of the reproduced image deteriorates significantly, especially if each signal on the screen has a different ratio from the preset ratio. It was possible.

The present invention was created in view of these problems, and its purpose is to provide a color image encoding device that takes into account both the amount of information held in a color image (detailed image) and the impact it has on human vision. Our goal is to provide the following.

[Structure of the invention]

(Means for Solving the Problems) The present invention provides a color image encoding device that encodes a color image.
It converts into G, B signals and luminance 9 color difference signals, calculates the amount of information of each of these signal components, sets the amount of code to be encoded or the ratio distribution based on the calculated amount of information, and calculates the amount of code to be encoded or the ratio of the amount of code to be distributed. This is a color image encoding device characterized in that signal components are encoded using a code amount or ratio and a preset code amount. It also includes a method of dividing one screen into blocks before converting a color image into signal components.

To explain in detail the case of blocking, the screen representing each R, G, B signal, luminance, and color difference signal of the input image is divided into blocks, and the variance, standard deviation, or pixels in the block are calculated for each block. By the absolute difference between the maximum and minimum values, or by the value of the output result when a filter is applied to the block (for example, a band pass filter that takes into account human visual characteristics), or alternatively.

Information about each block is determined by a part of the transform coefficients when each block is orthogonally transformed, or by the sum of the absolute values of all transform coefficients, power (some of the coefficients may be weighted in consideration of visual characteristics), etc. Calculate the amount. The average value of this information amount for one entire screen is taken as the information amount of each R, G, B signal or luminance two-color difference signal of that screen, and its ratio (for example, R
:G:B, luminance signal:color difference signal)
, B signal, or luminance, chrominance signal, or the ratio thereof (for example, here, [111
When Y = f (X) (f is a predetermined function), the distribution ratio of the number of coding bits of the luminance signal and the color difference signal is ).

During actual encoding, the number of encoded bits or its ratio determined as described above is used as a reference (methods for ensuring that the encoding process of the signal is completed within the number of encoded bits or its ratio, and the code Even if the encoding process exceeds the number of encoded bits or the ratio thereof, there may be a method in which the excess bits are processed within a predetermined range.) Encoding is performed.

Furthermore, within the color difference signal, for example, the RY signal and the B
-The impact it has on the human eye is different from the Y signal.
The RY signal is easier to detect than the B-Y signal. Therefore, when allocating the code amount, even within the color difference signal, R-Y
The signal is weighted more heavily than the B-Y signal.

Note that when calculating the amount of information for each signal above, it was defined as the average value of the values for each block, but regarding this amount of information, instead of calculating the value for each block, the entire screen is It is also possible to calculate the amount of information at once by considering it as a block.

(for production) In this way, each R, G, B signal or brightness.

The distribution of the number of bits encoded for each signal or its ratio is determined in advance from the ratio of the amount of information of the color difference signal, taking human visual characteristics into consideration, and during actual encoding, the allocated number of bits or its ratio is By encoding each signal based on the ratio, no matter what type of image is input, there will be no collapse on the screen, and moreover, it will always be possible to obtain a reproduced image with excellent visual characteristics. becomes possible.

(Example) The present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram of a transmitting side according to an embodiment of the present invention. First, the block division circuit 101 divides an input image (for example, Y, RY, and B-Y signals) into blocks (for example, 8×8 pixels or 16×16 pixels), and calculates the amount of information for each block. The amount of information is calculated in circuit 102.

Possible definitions of the amount of information include the variance (standard deviation) of pixel values within a block, the absolute value of the difference between the maximum pixel value and the minimum pixel value, and the like. The final information amount is calculated by averaging these for one screen. Further, in a weighting circuit 103, weighting is performed on the information amount of each signal in consideration of visual characteristics. For example, the information amount of the luminance signal is AY, and the information amount of the color difference signal is A.

(In this, the information amount of RY is AR-Y, and the information amount of BY is AB-y. In other words, A c = A R-Y
+ A B - Y), the weighted values are (IEAY, βAc (e.g. α: β
= 2: 1, 3: 1 @. Also, R-Y, B-Y
Considering the weighting for PAy T fAR−Y t rAB
-y (for example, p:y:r=10:3:2, etc.).

Also, a system without this weighting circuit 103 can be considered.In this case, the above-mentioned α, β, or Py
are all 1. Furthermore, the encoding bit allocation determining circuit 1
In step 04, the allocation of the number of encoded bits to each signal is determined based on the weighting information amount. In other words, the total code amount is BaJ
ゎThe code amount of the luminance 9 color differences (R-Y, B-Y) is set to B
Y,B.

If (BR-YI BB-y), The number of encoded bits of each signal is determined by .

Alternatively, the number of encoding bits for luminance and chrominance can be determined by (can also be determined individually).

On the other hand, as another determination method, it is also possible to use the ratio of information amounts. In other words, as shown in Figure 2, a straight line (or a curved line is fine)
Determine the information amount ratio of Y and C (AV/AC) on the horizontal axis,
The ratio of the number of encoded bits (By/Be) is plotted on the vertical axis, and by this straight line (or curve), from A y / A C to By
/Bc is uniquely determined (dotted line in the figure). Note that this straight line (or curve) may or may not pass through the origin. Further, the inclination is also determined in consideration of viewing characteristics, and may be changed depending on the pattern to be handled.

Further, when processing is performed using R, G, and B signals, similar processing can be performed, for example, R, G as weighting.

B=2:3:1 etc. can be considered.

Next, the encoding circuit 105 performs various types of encoding such as DCT (discrete cosine transform), VQ (vector quantization), and DPCM, or variable length encoding. here 10
The amount of information to be transmitted for each encoded signal is determined based on the number of encoded bits for each signal determined in step 4. For example, when encoding each signal of a Y signal, there is a method to ensure that the encoding process for that signal is completed within the determined number of encoding bits by once ending the encoding process within ]-, and the encoding process for that signal is Even when encoding is performed in excess of the number, there may be a method of processing such that the excess is within a predetermined range.

Note that although the number of encoding bits is determined in advance here, only the ratio of the number of encoding bits of each signal is determined, and that ratio is used as the standard (in other words, whether the standard is completely satisfied or the ratio is (make sure that it does not deviate too much from the
A method of performing encoding processing (that is, the number of encoding bits actually used changes, but the ratio is always constant or almost constant) is also considered.

Further, in this embodiment, block division is performed at step 101, but it is also possible to consider one screen of each signal as one block and calculate the amount of information at one time. In this case, block division will be performed in front of the encoding circuit 105 if necessary (or block division may not be necessary).

FIG. 3 is a block diagram of the transmitting side of another embodiment of the present invention. First, each signal (R, G, B or y, c) is divided into blocks in 201, and DCT is applied in a DCT circuit 202. Then, the information amount calculation circuit 203 calculates the amount of information for each block. In the case of the example, D within one block
The sum of absolute values, power, etc. is calculated for each coefficient value of the CT conversion coefficients. Furthermore, the weighting circuit 204 performs weighting in consideration of visual characteristics, and here, for example, as shown in FIG. (The midrange components are heavily weighted according to the In addition,
Although the amount of information may be determined from all the transform coefficients, it is also possible to use only a part of the transform coefficients. In that case, the weighting coefficient value of the unused area among the weighting coefficients shown in FIG. 4 may be set to φ.

Thereafter, the coding bit allocation determining circuit 205 determines the coding bit allocation or the coding bit allocation ratio, which is similar to that shown in the embodiment of FIG. Furthermore, in step 206, encoding processing after DCT is performed based on the determined number of encoding bits or the ratio thereof.

FIG. 5 is a block diagram of the transmitting side of another embodiment of the present invention. First, a block dividing circuit 401 divides the blocks into blocks, and then a filter 402 applies a filter (for example, in consideration of visual characteristics, there is a hand pass filter that passes mid-range frequency components).
filters, etc.). Thereafter, the information amount calculation circuit 403 calculates the information amount by performing the same processing as shown in the embodiment of FIG. 1 on each pixel value after applying the filter,
This is averaged over one screen to obtain the final information amount of each signal.

Thereafter, a coding bit allocation determining circuit 404 determines the number of coding bits (or a ratio of the number of coding bits) for each signal, and based on this, a coding circuit 405 performs coding processing.

Note that although weighting processing is not performed in this embodiment (because it is considered that the mid-range frequency is already weighted by passing through the hand bass filter), the weighting process is further performed after the information amount calculation circuit 403. It is also possible to attach.

In that case, the final amount of information for one screen will be determined based on the amount of information after weighting.

〔Effect of the invention〕

As explained above, according to the present invention, the signal components (each R, G
, B signal, or luminance two-color difference signal) and the impact each signal has on human vision are taken into account when encoding each signal. Since it is possible to allocate the number of encoded bits (or determine the ratio of the number of encoded bits), no matter what kind of image is input, there will be no collapse on the screen, and the visual characteristics will always be maintained. It becomes possible to obtain an excellent reproduced image.

[Brief explanation of the drawing]

Fig. 1 shows an embodiment of the present invention, Fig. 2 shows a luminance of 9
A diagram showing a straight line that determines the coding bit number allocation ratio from the information content of a color difference signal, FIG. 3 is a diagram showing another embodiment of the present invention, and FIG. 4 is a diagram showing the weighting of each coefficient in the DCT conversion plane. FIG. 5 is a diagram showing another embodiment of the present invention. 301...Block, 302...Low frequency component area, 303...Medium frequency component area, 304...
High frequency component region, 305...Weighting coefficient in region 304. Figure 1 Agent Patent attorney Nori Chika Ken Yudo Mitsuyuki Matsuyama AY/Ac Figure 2

Claims (6)

[Claims] (1) A color image encoding device that encodes a color image, comprising means for converting the color image into predetermined signal components, means for calculating the amount of information of each of the signal components obtained by this means, and and means for setting the distribution of the code amount or ratio to be encoded based on the information amount calculated by the means, and the code amount or ratio allocated by the means and the code amount set in advance are used to encode the signal. A color image encoding device characterized by encoding components. (2) The color image encoding device according to claim 1, wherein the means for calculating the amount of information includes means for performing weighting processing in consideration of visual characteristics for each signal component. (3) In a color image encoding device that encodes a color image, means for dividing one screen of the color image into blocks;
means for converting each block divided by this means into predetermined signal components; means for calculating the amount of information of each of the signal components obtained by this means; and a code based on the amount of information calculated by this means. means for setting the distribution of the amount of code or ratio to be encoded, and the signal component of each block is encoded using the amount of code or ratio allocated by this means and the amount of code set in advance. Color image encoding device. (4) The means for calculating the amount of information is determined by the average value for one screen of statistics representing the fineness of the picture in the block obtained from the pixel values in each block where one screen is divided into blocks. The color image encoding device according to claim 3, characterized in that: (5) The means to calculate the amount of information is to divide one screen into blocks, apply orthogonal transformation to each block, and then express the fineness of the picture in the block obtained from some or all of the transformation coefficients. 4. The color image encoding apparatus according to claim 3, wherein the determination is made based on an average value of the statistics in one screen. (6) The color image encoding device according to claim 3, wherein the means for calculating the amount of information includes means for performing weighting processing in consideration of visual characteristics for each signal component.