RU2336663C1 - Method and device of colour image coding/decoding using chroma signal components correlation - Google Patents

Abstract

FIELD: physics.

SUBSTANCE: invention concerns image processing technology, particularly YCbCr-format colour image data coding/decoding to smaller data volume by finding correlation between Cb and Cr chroma signal components of colour image data. The invention claims colour image coding method involving stages of: chroma signal component conversion in each of two or more mutual prediction modes; cost calculation for conversion values in each of two or more mutual prediction modes with the help of cost function defined preliminarily; selection of one or more mutual prediction modes on the basis of calculation result and conversion value output for the selected mutual prediction mode; entropic coding of output conversion values, where preliminarily defined cost function is selected out of cost function defining distortion in dependence of transfer rate, function of absolute subtract value amount, function of absolute converted subtract, function of square subtract sum and function of average absolute subtract.

EFFECT: increased efficiency of image coding.

88 cl, 23 dwg

Description

FIELD OF THE INVENTION

This application claims the priority of Patent applications (Korea), numbers 10-2004-0116962 and 10-2005-0027827, filed December 30, 2004 and April 2, 2005, respectively, in the Korean Intellectual Property Office, the essence of which is fully contained in this document link.

The devices and methods in accordance with the present invention are associated with the encoding and decoding of color image data, and more particularly, with the encoding and decoding of color image data having the YCbCr format to a smaller amount of data by finding a correlation between the components of the color signal Cb and Cr data color images.

State of the art

1 is a diagram illustrating data constituting a video having an RGB format and video having a YCbCr format.

The RGB format represents color video, divides the color component of the color video signal into red (R), green (G) and the blue component (B) of the color signal and represents the components of the color signal R, G and B. Moreover, the components of the color signal R, G and B have the same amount of data. For example, when a macroblock has a size of 16 × 16, the color components R, G, and B are 16 × 16. However, the human eye is more sensitive to luminance components representing luminance than to chroma components representing colors. Thus, a format in which color video is divided into luminance and chrominance signal components to be presented can be used to reduce the amount of data. YCbCr format is such a format.

In the YCbCr format, more data is allocated to luminance components than to chroma components. Referring to FIG. 1, when an RGB format video with a 16 × 16 macroblock is represented as a YCbCr format video in a 16 × 16 macroblock, an RGB format video is represented as a 16 × 16 luminance signal block and 8 × 8 Cb and Cr color signal blocks . In this case, the values of the component of the luminance signal Y and the components of the chrominance signal Cb and Cr are calculated by weighted combinations of the values of R, G and B. For example, the values of the component of the luminance signal Y and the components of the chrominance signal Cb and Cr are calculated using equations such as Y = 0, 29900R + 0.58700G + 0.11400B, Cb = -0.16874R-0.33126G + 0.500000B and Cr = 0.500000R-0.41869G-0.08131B. As described above, color film data having a YCbCr format includes a luminance signal component and two color signal components. When the color cinema image data is encoded, the luminance component and the two color components are encoded separately. In other words, the luminance component and the two chroma components are encoded independently of the correlation between the two chroma components.

Figure 2 is a diagram of the data structures of color video in the formats 4: 4: 4, 4: 2: 2 and 4: 2: 0.

When a motion picture is encoded, the color format of the motion picture is represented by the transmission speed of the luminance signal component and the chrominance components of the movie pixel in the horizontal pixel line. Further, the luminance signal component is denoted by Y, and the color signal components are denoted by Cb and Cr. The luminance signal (luminance) of one pixel is represented using eight bits in the ITU-R Recommendation, and the chroma signal (color) of a pixel is represented using the two components of the chrominance signal Cb and Cr, each of which has eight bits. The coordinate system for representing colors is called the color space. In the standards of the Expert Group on Cinematography (MPEG), the color format of a movie image is represented using three 8-bit pieces of information, i.e. component of the luminance signal Y and chrominance components Cb and Cr.

When a color motion picture is represented using the luminance component Y and the color components Cb and Cr, several types of color formats can be represented according to the transmission speed of the luminance component Y and the color components Cb and Cr. In the case of different color formats, the components of the luminance signal Y of the different color formats are the same, but the components of the color signal Cb and Cr of the different color formats vary. Referring to FIG. 2, a video having a 4: 2: 2 format is obtained by down-sampling by half the color signal components of a video having a 4: 4: 4 format in the horizontal direction, and a video having a 4: 2: 0 format , obtained by sampling with decreasing the frequency of half the components of the color signal of a video having a 4: 2: format in the vertical direction.

As described above, in a conventional codec (MPEG, H.26x, VC1), an RGB color video is converted to a YCbCr color video to separate the luminance signal component and the color signal components from the YCbCr color video, so as to separately encode the luminance signal component and the color signal components . At the same time, color video can have several different formats, for example, 4: 4: 4, 4: 2: 2 and 4: 2: 0 formats, etc. In general, a traditional codec (MPEG, H.26x, VC1) receives video data having a 4: 2: 0 format to encode the component of the luminance signal Y and the components of the chrominance signal Cb and Cr. The following describes an example of video data having a 4: 2: 0 format.

SUMMARY OF THE INVENTION

Technical challenge

In a general movie coding method, the component of the luminance signal Y and the components of the chrominance signal Cb and Cr are encoded so as not to have temporal and spatial redundancies. Spatial redundancy is removed by intra-prediction between the neighboring block and the current block, and temporal redundancy is removed by inter-prediction between the previous image and the current image. Moreover, only the difference component between the neighboring block and the current block and only the difference component between the previous image and the current image are encoded by internal prediction, so as to increase the compression efficiency.

In other words, only predictions are performed to remove temporal and spatial redundancies of the component of the luminance signal Y and the components of the chrominance signal Cb and Cr. The removal of redundancy by correlation between the component of the luminance signal Y and the components of the chrominance signal Cb and Cr is not performed. However, when compressing high-quality video, for example, advanced H.264, the data volume of the component of the luminance signal Y and the components of the chrominance signal Cb and Cr increases. Thus, a method is needed for efficiently compressing high quality video.

Technical solution

The present invention provides an apparatus and method for encoding and decoding color images by which a correlation between the components of the color signal Cb and Cb of a color image is found and used in order to reduce the amount of data to be encoded in order to increase the encoding rate.

According to an aspect of the present invention, there is provided an encoding apparatus including: a color signal component converter, multiplying the color signal components Cb and Cr of a color video by predetermined coefficients, adding up the multiplication results to form a plurality of conversion values, selecting two conversion values at the lowest cost calculated by a predetermined cost function and outputting the selected conversion values; and an entropic encoder performing entropic encoding of the selected transform values.

The components of the chrominance signal Cb and Cr can be transformed and quantized components of the chrominance signal. The encoding device may further include a converter and a quantizer, converting and quantizing conversion values output from the color component converter, if the color component Cb and Cr are not converted or non-quantized color component.

The color signal component converter can calculate the conversion values of the color signal components Cb and Cr using the equation: conversion value = ax Cb + bx Cr + c, where a, b and c are constants, and many sets (a, b, c) are preliminarily user defined. A predetermined cost function can be one of the cost functions that determines the distortion depending on the transmission speed, the sum function of the absolute difference values, the sum function of the absolute transformed difference, the sum function of the difference square and the mean absolute difference function.

The converter of the components of the chroma signal may include: a transmitter of conversion values multiplying the components of the chroma signal Cb and Cr by a plurality of coefficients (a, b, c), summing the multiplication results and generating the conversion values; a cost calculator that calculates the costs of the conversion values using a predefined cost function; and a determinant that selects and outputs the two conversion values that have the least cost.

The converter of the components of the color signal can encode information according to the coefficients (a, b, c) corresponding to the two conversion values that have the least cost.

According to another aspect of the present invention, there is provided an encoding method comprising the steps of: multiplying the components of the color signal Cb and Cr of a color video by predetermined coefficients, summing the multiplication results to form a plurality of conversion values, selecting two conversion values having the least cost calculated by a predetermined cost function and outputting selected conversion values; and perform entropic encoding of the selected transform values.

Multiplying the components of the color signal Cb and Cr of the color video by predetermined coefficients, summing the multiplication results in order to generate a plurality of conversion values, selecting two conversion values having the lowest costs calculated by a predetermined cost function, and outputting the selected conversion values may include stages in which: multiply the components of the color signal Cb and Cr by a set of coefficients (a, b, c), summarize the results of multiplication and miruyut conversion values; calculating the costs of the conversion values using a predefined cost function; and select and output two conversion values that have the least cost.

Information on the coefficients (a, b, c) corresponding to the two least cost conversion values may be group encoded.

According to yet another aspect of the present invention, there is provided an encoding apparatus including: an entropic decoder performing entropic decoding of an encoded bitstream; and a color component inverter that ignores the decoded data if the decoded data is a component of the luminance signal and extracts information on the coefficients by which the color components Cb and Cr are multiplied and added to form and output the original color components Cb and Cr if decoded data are the components of the color signal.

The chroma component inverter can extract information indicating which set of coefficients (a, b, c) is used to encode the chroma components to calculate the chroma components Cb and Cr, the information being group encoded and transmitted.

According to yet another aspect of the present invention, there is provided an encoding method including the steps of: performing entropic decoding of an encoded bitstream; and ignoring the decoded data if the decoded data is a component of the luminance signal, and extracting information about the coefficients by which the components of the color signal Cb and Cr are multiplied and added to form the output source components of the color signal Cb and Cr, if the decoded data is components of the color signal.

According to yet another aspect of the present invention, there is provided a color image encoding apparatus including: a color signal component converter, converting color signal components of a color image in each of two or more mutual prediction modes, calculating costs for conversion values in each of two or more mutual prediction modes using a predefined cost function that selects one of two or more mutual forecast modes ozirovanie on the basis of the calculation result and outputting the conversion values of the selected mutual prediction mode; and an entropic encoder performing entropic encoding of the output transform values.

The selection of one of two or more mutual prediction modes may be performed in a group segment of a predetermined macroblock. Moreover, information about the mutual prediction mode selected for a predefined macroblock can be encoded in a segment of a predefined group containing many blocks.

Mutual prediction mode information for a plurality of blocks of a predetermined group can be classified into a plurality of mode matrices, and a plurality of mode matrices are encoded.

The plurality of mode matrices may include information on whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks.

A predetermined mode matrix can be obtained by assigning the mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is set to 1, and assigning the mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is not applied , the values are '0'.

The mutual prediction mode information for a plurality of blocks of a predetermined group can be classified into a plurality of mode matrices in each mode, a plurality of mode matrices can be placed in a specific order, information about the next mode matrix can be converted based on the mode information about the previous mode matrix, and the transformed matrix mode can be encoded.

The plurality of mode matrices may include information about whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks, and the next mode matrix can be transformed by deleting information about the block to which the mutual prediction mode is applied to the previous mode matrix , from information about the next mode matrix based on information about the previous mode matrix.

A predetermined mode matrix can be obtained by assigning the mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is set to 1, and assigning the mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is not applied , the values are '0'.

Deletion of information about the block to which the mutual prediction mode is applied to the previous mode matrix can be done by assigning the information about the block to the value '0'.

Many blocks can be macroblocks, and a predefined group can be an image.

The Converter of the components of the color signal may include: a repository of tables of mutual prediction modes, storing a table of mutual prediction modes, containing two or more modes of mutual prediction; a conversion value calculator calculating conversion values of the color signal components Cb and Cr of the color image in each mode based on a table of mutual prediction modes; and a mode selector selecting a mutual prediction mode in which the conversion values have the lowest costs calculated by a predetermined cost function.

The encoding device may further include a group coding encoder, an RLE encoder that executes information about the selected mutual prediction mode.

According to yet another aspect of the present invention, there is provided a method for encoding a color image, comprising the steps of: converting the color signal components of a color image in each of two or more mutual prediction modes, calculating costs for conversion values in each of two or more modes forecasting using a predefined cost function, choose one of two or more modes of mutual forecasting based on the calculation result and outputting conversion values of the selected mutual prediction mode; and perform entropic coding of the output transform values.

According to yet another aspect of the present invention, there is provided a device for decoding an encoded color image, including: an entropic decoder performing entropic decoding of an input bit stream; and an inverse component of the chroma signal, restoring the original components of the chroma signal based on the information of the mutual prediction mode applied to the current block having a predetermined size, and the information of the mutual prediction mode is extracted from the input bit stream. At the same time, the information of the mutual prediction mode may indicate the mutual prediction mode of two or more mutual prediction modes applied to the current block, and the original components of the color signal can be obtained from the conversion values corresponding to the mutual prediction mode applied to the current block.

According to yet another aspect of the present invention, there is provided a method for decoding a color image, comprising the steps of: performing entropic decoding of an input bit stream; and restoring the original components of the color signal based on the information of the mutual prediction mode applied to the current block having a predetermined size, and the information of the mutual prediction mode is extracted from the input bit stream. At the same time, the information of the mutual prediction mode may indicate the mutual prediction mode of two or more mutual prediction modes applied to the current block, and the original components of the color signal can be obtained from the conversion values corresponding to the mutual prediction mode applied to the current block.

According to yet another aspect of the present invention, there is provided a computer-readable recording medium having a computer program embodied therein for an encoding method.

According to yet another aspect of the present invention, there is provided a computer-readable recording medium having a computer program embodied therein for a decoding method according to the claims.

Benefits

According to the present invention, the compression efficiency of motion pictures can be improved, and the number of bits required for encoding can be significantly reduced, and efficient group encoding can be performed.

Description of drawings

The above and / or other aspects of the present invention should become more apparent by describing exemplary embodiments with reference to the accompanying drawings, of which:

Figure 1 is a diagram of the data constituting the video having the RGB format and the video having the YCbCr format;

FIG. 2 is a diagram of video data structures having 4: 4: 4, 4: 2: 2 and 4: 2: 0 formats;

3 is a block diagram of a motion picture encoding apparatus according to an exemplary embodiment of the present invention;

4 is a diagram illustrating the calculation of the conversion values of the components of the color signal according to an exemplary embodiment of the present invention;

FIG. 5 is a block diagram of a converter 330 of the color signal components shown in FIG. 3;

6 is a flowchart of an encoding method according to an exemplary embodiment of the present invention;

7 is a block diagram of a decoding apparatus according to an exemplary embodiment of the present invention;

FIG. 8 is a flowchart of a decoding method according to an exemplary embodiment of the present invention; FIG.

9 is a table showing a mutual prediction mode according to an exemplary embodiment of the present invention;

10 is a table illustrating a reverse mutual prediction method with respect to each mutual prediction mode;

11 is a block diagram of a converter 330 of the color signal components shown in FIG. 3 according to an exemplary embodiment of the present invention;

12 is a block diagram of a mutual prediction mode selected for each macroblock in one image;

13A-13E are views showing matrices of mutual prediction modes according to an exemplary embodiment of the present invention;

14A-14D are views illustrating a method of encoding mutual prediction mode information according to an exemplary embodiment of the present invention;

FIG. 15 is a flowchart of a method for encoding mutual prediction mode information according to an exemplary embodiment of the present invention; FIG. and

16 is a flowchart of a decoding method according to an exemplary embodiment of the present invention.

Optimal mode

MODE FOR CARRYING OUT THE INVENTION

In the following, exemplary embodiments of the present invention are described in detail with reference to the accompanying drawings.

3 is a block diagram of a motion picture encoding apparatus according to an exemplary embodiment of the present invention. The device includes a motion estimation unit 302, a motion compensation unit 304, an intra prediction unit 306, a converter 308, a quantizer 310, a reconfiguration unit 312, an entropic encoder 314, an inverse quantizer 316, an inverse converter 318, a filter 320, and a frame memory 322.

A device encodes macroblocks of a current image in one of a plurality of encoding modes. For this purpose, the device performs coding in modes in which mutual prediction and intra prediction are performed in order to calculate a cost function that determines the distortion depending on the transmission rate (RDCosts). The device determines the mode in which the lowest RDCosts value is calculated as the optimal mode and performs encoding in the optimal mode. The bit rate (R) indicates the bit rate, which is the number of bits used to encode one macroblock. Specifically, R is a value obtained by adding the number of bits obtained by encoding the residual signal generated after the mutual prediction and intra prediction are performed and the number of bits obtained by encoding the motion vector. Distortion (D) indicates the difference between the original video macroblock and the decoded video macroblock. Thus, D is the value obtained by decoding the original macroblock.

However, determining the optimal encoding mode can be performed using various methods, as well as calculating RDCosts. In other words, the calculation of RDCosts and other costs can also be performed using various methods. For example, examples of a suitable cost function include the sum of absolute difference values (SAD), the sum of the absolute transformed difference (SATD), the sum of the square of the difference (SSD), the mean of the absolute difference (MAD), the Lagrange function, and the like.

For mutual prediction, the motion estimation unit 302 searches the estimated image for the macroblock of the current image in the reference image. The motion compensation unit 304 calculates an intermediate value of the pixels of the reference blocks found in pixel units 1/2 or 1/4 to determine the value of the data of the reference blocks. Therefore, the motion estimation unit 302 and the motion compensation unit 304 perform mutual prediction.

The intra prediction unit 306 performs intra prediction to search the current image for the estimated macroblock value of the current image. A determination is made as to whether mutual prediction or intra prediction is performed with respect to the current macroblock by determining a coding mode in which the smallest RDCosts value is calculated as the coding mode of the current macroblock so as to encode the current macroblock.

As described above, if the estimation data on which the macroblock of the current frame should rely is found by performing mutual prediction or intra prediction, the estimation data is subtracted from the macroblock of the current image, and then the subtraction result is input to the color component converter 330. The color signal component converter 330 receives the color signal components, converts the signal components in color into different conversion values according to the method for converting the color signal components, which is described below, and selects two of the plurality of conversion values. In the case where the chroma component converter 330 receives the luminance component, the chroma component converter 330 passes the luminance component. The luminance component or the selected chroma components are inputted and converted by the converter 308 and then quantized by the quantizer 310. The result of subtracting the reference block with motion estimation from the macroblock of the current frame is called the remainder. The data input to the color component converter 330 to reduce the amount of data during encoding is a residual value. The quantized residual value passes through the reconfiguration unit 312 and is encoded by the entropic encoder 314.

The quantized image is decoded into the current image by the inverse quantizer 316 and the inverse transformer 318 to obtain a reference image, which should be used for mutual prediction. The decoded current image is stored in frame memory for use in inter-prediction of the next image. When the decoded image passes through the filter 320, the decoded image becomes the original image, including a small number of coding errors.

The following describes in detail the operation of the Converter 330 components of the color signal. When the color component converter 330 receives the color components Cb and Cr, the color component converter 330 calculates the conversion value using equation 1:

Conversion value = a x Cb + b x Cr + c ... (1)

- in this case, a, b and c can be determined experimentally. For example, if (a, b, c) is (1, 0, 0), (0, 1, 0), (-1, 1, 0) or (1, 1, 0), the conversion value is Cb , Cr, —Cb + Cr or Cb + Cr. Costs for Cb, Cr, -Cb + Cr and Cb + Cr are calculated. Calculation of costs and used cost functions - such as described above. The two sets (a, b, c) having the lowest value of the calculated costs are selected and then input to the transformer 308. For example, if Cb and -Cb + Cr are selected, the transformer 308 converts the Cb and -Cb + Cr components. In this case, the costs are the smallest. Thus, the values of Cb and -Cb + Cr are the smallest, and thus, the bit rate required for encoding is small. In the case of a macroblock for internal forecasting, (a, b, c) can be equal to (-1, 1, 14), (1, 1, -250), (1, 0, 14) or (0, 1, 14). Even in the case of a macroblock, internal prediction calculates the costs for a pair of coefficients (a, b, c), and the components of the color signal, determined by (a, b, c), for which the costs are the smallest, are found and encoded. The converter 330 of the chroma signal components may be placed after the converter 308 and the quantizer 310. That is, costs are calculated using the converted chrominance components Cb and Cr in the frequency domain rather than in the spatial domain to perform reconfiguration and entropic coding.

4 is a diagram illustrating the calculation of conversion values of the components of a color signal according to an exemplary embodiment of the present invention. Referring to FIG. 4, a single pixel value is read from each of the blocks Cb and Cr and multiplied by or added to (a, b, c) using equation 1 above to calculate the conversion values.

FIG. 5 is a block diagram of a converter 330 of the color signal components shown in FIG. 3.

The converter 330 of the components of the color signal includes a transmitter 510 of the conversion values, the calculator 520 costs and block 530 determination. When the chrominance components Cb and Cr are input, the transform value calculator 510 calculates the transform values of all cases that can be obtained from the sets of coefficients (a, b, c) using equation 1 above. Cost calculator 520 calculates costs for conversion values. The determination unit 530 selects the two lowest cost values to output conversion values having the two lowest cost values.

6 is a flowchart of an encoding method according to an exemplary embodiment of the present invention.

When video data is inputted, in steps S610 and S620, motion estimation and motion prediction are performed in the case of mutual prediction. In the case of internal forecasting, motion estimation and motion prediction are omitted. In this case, motion estimation and motion prediction are performed as described with reference to FIG. 3. At step S630, the costs of all cases are calculated using the predefined coefficients (a, b, c), as described above with reference to FIG. 4 and 5. In step S640, the two lowest cost values are selected. In steps S650, S660, and S670, the two smallest values are converted, quantized, and entropically encoded, respectively. Conventionally, chrominance components Cb and Cr are encoded using this method. However, in the present invention, in steps S630 and S640, the redundancy between the color components Cb and Cr is removed before encoding in order to reduce the number of bits required for encoding.

The information of the selected (a, b, c) coefficients is encoded and transmitted. Information of the selected coefficients for each macroblock is recorded in the image header to indicate which components of the color signal of each macroblock are encoded and transmitted. In the above mutual prediction, if the first and third coefficients are selected from the coefficients (1, 0, 0), (0, 1, 0), (-1, 1, 0) and (1, 1, 0), group coding is performed for first and third coefficients.

In more detail, group coding is performed for the information of the selected coefficients only when a block of chroma components is encoded. In this case, a chroma coding block scheme with a conventional syntax or a coding block scheme for a chroma signal (CBPC) may be used. When batch coding is performed, the number of bits allocated to the “series” varies depending on how many sets are used, and the number of bits allocated to the “length” varies depending on how many continuous series are encoded into one set. For example, in the case when the number of sets is four, i.e. S1, S2, S3 and S4, the number of bits allocated to the “series” is two bits, and the number of bits allocated to the “length” is five bits, one set (series, length) is encoded into seven bits. Thus, if S1 is continuously outputted eleven times, (S1, 10) is encoded in 0001010. Since the information of the selected coefficients of each macroblock is very likely to have a value similar to the information of the coefficients of the neighboring macroblock, group coding can be used to reduce the number of bits, required for coding. In addition, by using a chroma signal coding block scheme or CBPC, information is transmitted to indicate whether a chroma signal block is encoded in segments of a group of blocks.

Information of the coefficients (a, b, c) selected for each macroblock is likely to be similar for the neighboring macroblock. Thus, group encoding can be used to reduce the number of bits required for encoding.

7 is a block diagram of a decoding apparatus according to an exemplary embodiment of the present invention. Referring to FIG. 7, a decoding device includes an entropic encoder 702, a reconfiguration unit 704, an inverse quantizer 706, an inverse converter 708, an inverse color converter 710, a motion compensation unit 712, an intra prediction unit 714, an intra prediction unit 714, a filter 716 and frame memory 718 . When the encoded bit stream is input to the decoding device, the encoded bit stream is entropically decoded, reconfigured, converted back, and input to the inverse color component converter 710. In the case where the input is the luminance component, the luminance component is ignored. In the case where the input is color component, the inverse color component converter 710 determines which coefficients (a, b, c) were used to encode the color component, thereby forming the color component Cb and Cr. Information indicating which coefficients (a, b, c) were used to encode and transmit the components of the color signal is plagued by group coding and transmitted. Thus, the inverter 710 components of the chroma signal decodes the information to form the components of the chrominance signal Cb and Cr. Alternatively, an inverter 710 of the chrominance components may also be placed before inverse quantizer 706 and inverse converter 708.

Fig. 8 is a flowchart of a decoding method according to an exemplary embodiment of the present invention.

At step S810, entropic decoding is performed. In step S820, inverse quantization is performed. At step S830, the inverse transform is performed. In step S840, the received coefficient information (a, b, c) is decoded, a determination is made as to which combination of the chrominance components Cb and Cr is encoded to perform inverse transform of the combination and obtain chrominance components Cb and Cr. At step S850, motion compensation is performed. In the case of internal forecasting, motion compensation is omitted.

The following describes the encoding method according to an exemplary embodiment with reference to FIGS. 9-15.

9 is a table showing a mutual prediction mode according to an exemplary embodiment of the present invention. Referring to FIG. 9, five mutual prediction modes, 0-4, are set with respect to each macroblock. One of the mutual prediction modes is selected in the macroblock group segment, as shown in FIG. Blocks Cb and Cr are replaced with conversion values 1 and 2 according to the selected mode and table shown in Fig.9. The correlations between conversion values and Cb and Cr values in the mutual prediction modes shown in FIG. 9 are exemplary. Alternatively, modes having other correlations may be added, or modes may be deleted.

For example, if the mutual prediction mode selected with respect to the predetermined macroblock is 0, the transform value 1 is the Cb value of the Cb block, and the transform value 2 is the Cr value of the Cr block. In addition, if the selected mutual prediction mode is 1, the transform value 1 is the Cb value of the Cb block, and the transform value 2 is the value obtained by subtracting the Cr value of the Cr block from Cb ', which is the restored Cb value. Using the reconstructed Cb value and the Cr value in the mutual prediction mode shown in FIG. 9 serves to further accurately decode the Cb and Cr values. Alternatively, instead of the reconstructed Cb and Cr values, the original Cb and Cr values may be used. In this case, the reconstructed values of Cb and Cr are obtained by converting and quantizing the original values of Cb and Cr and then back quantizing and inverting the original values of Cb and Cr. The initial values of Cb and Cr are the values of Cb and Cr that are not converted and quantized.

10 is a table illustrating a reverse mutual prediction mode with respect to a mutual prediction mode. Referring to FIG. 10, if the inverse mutual prediction mode for one macroblock is 0, the values of Cb and Cr of the corresponding macroblock are obtained from transform values 1 and 2. If the inverse mutual prediction mode is 1, the Cb value of the corresponding macroblock is obtained from the transform value 1, but the Cr value is obtained from the value obtained by subtracting the conversion value 2 (Cb'-Cr) from the restored value Cb 'of the conversion value 1. In this case, subtracting the conversion value 2 and h restored value Cb 'due to the fact that the value of conversion 2 includes Cb'.

FIG. 11 is a block diagram of a color signal converter 330 of FIG. 3 according to an exemplary embodiment of the present invention.

The converter 330 of the chroma signal components includes a transform value calculator 1110, a mutual prediction mode table storage 1112, a cost calculator 1120, a mode selection and output value conversion unit 1130, and a mode selection storage 1132.

When the chrominance components Cb and Cr are input, the transform value calculator 1110 calculates the transform values 1 and 2 for each mutual prediction mode stored in the mutual prediction mode table storage 1112. For example, transform value calculator 1110 generates transform values 1 and 2 with respect to each mutual prediction mode shown in FIG. 9. In addition, in the present embodiment, the mutual prediction mode table is stored in the transform value calculator 1110 and the mutual prediction mode table 1112. Alternatively, the mutual prediction mode table may be stored in a predetermined location of the mutual prediction mode calculator 1110.

The calculator 1120 costs calculates the costs in relation to the conversion values calculated in each mode of mutual forecasting.

Block 1130 select the mode and output the conversion values selects a mutual prediction mode in which the conversion values are the least cost, and displays the conversion values. For example, if the mutual prediction mode 1 is selected according to the mutual prediction mode table shown in FIG. 9, the mode selection and conversion value output unit 1130 outputs the values of Cb and Cb'-Cr as the transformation values 1 and 2.

The mode selection store 1132 stores mode information for each macroblock selected by the mode selection unit 1130 and outputting conversion values. The mode information for each macroblock stored in the mode selection storage 1132 is used to form a table of mutual prediction modes in a segment of the image group shown in FIG. 12. In addition, in the present exemplary embodiment, the mode information for each macroblock is stored in the mode selection storage 1132. However, the mode information for each macroblock may be stored in a predetermined location of the mode selection unit 1132 and outputting conversion values.

The conversion values output from the mode selection unit 1130 and the output of the conversion values are output to a converter 308 and a quantizer 310 so as to be converted and quantized.

As described above, the conversion value is determined depending on the selected mode. Encoding is performed depending on a specific conversion value. Thus, the decoder should be informed which mutual prediction mode is selected for each macroblock. The following describes a method for transmitting information of the mutual prediction mode selected for each macroblock, with reference to FIGS. 12-14.

12 is a view illustrating a mutual prediction mode selected for each macroblock in one image. The value in each position is 0, 1, 2, 3, or 4, which indicates the mutual prediction mode applied to each macroblock corresponding to each position. For example, the value '0' at the top and leftmost position indicates that the Cb and Cr values of the corresponding macroblock are replaced according to the mutual prediction mode 0 in the table shown in Fig. 9, i.e. conversion value 1 is Cb, and conversion value 2 is Cr. The values 2 and 2 corresponding to the macroblocks next to the macroblock in the uppermost and leftmost positions show that the Cb and Cr values of the corresponding macroblocks are replaced according to the mutual prediction mode 2 shown in FIG. 9, i.e. the conversion value 1 is Cb, and the conversion value 2 is Cb '+ Cr.

13A-13E are views showing the value of the mutual prediction mode of each macroblock shown in FIG. 12 in the matrix of each mutual prediction mode.

FIG. 13A illustrates a mode 0 matrix that is reconfigured so that a macroblock indicating a mutual prediction mode 0 in the mutual prediction mode table shown in FIG. 12 is set to “1”, and other macroblocks not indicating a mutual prediction mode 0, have a value of '0'. For example, the values of the first, fourth, fifth, seventh, eighth, tenth and fourteenth macroblocks having the mutual prediction mode value 0 at the highest position are set to '1', and the values of other macroblocks at the highest position are set to '0'.

FIG. 13B illustrates a mode 1 matrix that is reconfigured so that a macroblock indicating the mutual prediction mode 2 in the mutual prediction mode table shown in FIG. 12 is set to “1”, and other macroblocks not indicating the mutual prediction mode 0, have a value of '0'. For example, the values of the sixth and ninth macroblocks having the mutual prediction mode value '1' in the uppermost position are set to '1', and the values of other macroblocks in the uppermost position are set to '0'.

FIG. 13C illustrates a mode 2 matrix that is reconfigured so that a macroblock indicating the mutual prediction mode 2 in the mutual prediction mode table shown in FIG. 12 is set to “1” and other macroblocks not indicating the mutual prediction mode 0, have a value of '0'. For example, the values of the second, third, thirteenth, fifteenth, seventeenth, nineteenth, twentieth, twenty first and twenty second macroblocks having the value of the mutual prediction mode 2 at the highest position are set to '1', and the values of other macroblocks at the highest position are set equal to '0'.

Fig.13D illustrates a matrix of mode 3, which is reconfigured so that the macroblock indicating the mutual prediction mode 3 in the table of mutual predictions shown in Fig, has the value '1', and other macroblocks that do not indicate the mutual prediction mode 0, have a value of '0'. For example, there are no macroblocks that have a mutual prediction mode 3 value in the highest position. Thus, all macroblocks are assigned the value '0'.

FIG. 13E illustrates a mode 4 matrix that is reconfigured so that a macroblock indicating the mutual prediction mode 4 in the mutual prediction mode table shown in FIG. 12 is set to “1” and other macroblocks not indicating the mutual prediction mode 0, have a value of '0'. For example, the values of the eleventh, twelfth, and sixteenth macroblocks having the mutual prediction mode 4 in the uppermost position are set to '1', and the values of the other macroblocks in the uppermost position are set to '0'.

In the case where the mutual prediction mode table shown in FIG. 12 is divided into mode matrices as shown in FIGS. 13A-13E, the length of the series 0 becomes longer.

14A-14D are views illustrating a method of encoding mutual prediction mode information according to an exemplary embodiment of the present invention. In other words, FIGS. 14A-14D show mode matrices that are transformed such that the lengths of one series of mode matrices shown in FIGS. 13A-13E become longer using the mode matrix reduction circuit of the present invention.

Fig. 14A shows a transformed matrix of mode 1 in which the values '0' corresponding to macroblocks having the value of the mutual prediction mode '1' in the matrix of mode 0 shown in Fig. 13A are removed from the matrix of mode 1 shown in Fig. 13B. As shown in FIG. 14A, 22 bits, 0000010010000000000000, at the uppermost position in the mode 1 matrix shown in FIG. 13B are converted to 15 bits, 001100000000000, of which the values are '0' of the first, fourth, fifth, seventh, eighth, the tenth and fourteenth macroblocks having the mutual prediction mode value '1' in the mode 0 matrix are deleted. The converted mode 1 matrix has a lower bit rate and a longer series than the mode 1 matrix.

FIG. 14B shows a transformed mode 2 matrix in which values '0' corresponding to macroblocks having the mutual prediction mode value '1' in mode 0 matrix and mode 1 matrix are removed from mode 2 matrix shown in Fig. 13C. As shown in FIG. 14B, 22 bits, 0110000000001010111111, at the highest position in the mode 2 matrix shown in FIG. 13C are converted to 13 bits, 1100110111111, of which the values are '0' of the first, fourth, fifth, seventh, eighth, the tenth and fourteenth macroblocks having the mutual prediction mode value '1' in the mode matrix 0, and the values '0' of the sixth and ninth macroblocks having the mutual prediction mode values '1' in the mode 1 matrix, are deleted. The converted mode 2 matrix has a lower bit rate and a longer series than the mode 2 matrix.

Fig. 14C shows a transformed mode 3 matrix in which values of '0' corresponding to macroblocks having the mutual prediction mode value '1' in mode 0 matrix, mode 1 matrix and mode 2 matrix are removed from mode 3 matrix shown in FIG. 13D. As shown in FIG. 14C, 22 bits, 0000000000000000000000, in the uppermost position in the mode 3 matrix shown in FIG. 13D, are converted to 3 bits, 000, of which the '0' values of the first, fourth, fifth, seventh, eighth, tenth and fourteenth macroblocks having the mutual prediction mode value '1' in the mode matrix 0, values '0' of the sixth and ninth macroblocks having the mutual forecasting mode values '1' in the mode matrix 1, and values '0' of the second, third, thirteenth , fifteenth, seventeenth, eighteenth, nineteenth, twenty first, twenty-first, and twenty second macroblocks having the inter-prediction mode value "1" on the mode 2 matrix removed. The converted mode 3 matrix has a lower bit rate and a longer series than the mode 3 matrix.

Fig.14D shows a transformed matrix of mode 4, in which the values '0' corresponding to macroblocks having the value of the mutual prediction mode '1' in the matrix of mode 0, matrix of mode 1, matrix of mode 2 and matrix of mode 3 are removed from the matrix of mode 4, shown in fig.13E. As shown in FIG. 14D, 22 bits, 0000000000110001000000, in the uppermost position in the mode 4 matrix shown in FIG. 13D, are converted to 3 bits, 111, of which the '0' values of the first, fourth, fifth, seventh, eighth, tenth and fourteenth macroblocks having the mutual prediction mode value '1' in the mode matrix 0, values '0' of the sixth and ninth macroblocks having the mutual forecasting mode values '1' in the mode matrix 1, and values '0' of the second, third, thirteenth , fifteenth, seventeenth, eighteenth, nineteenth, twenty first, twenty-first, and twenty second macroblocks having the inter-prediction mode value "1" on the mode 2 matrix removed. The converted mode 4 matrix has a lower bit rate and a longer series than the mode 4 matrix. In addition, all values on the converted mode 4 matrix have a value of '1'. Thus, encoding on the transformed matrix of mode 4 is not required.

In the present exemplary embodiment, the mode 0 matrix and the transformed matrixes of modes 1, 2, 3, and 4 having long series lengths 1 may be group encoded and then transmitted. Thus, the amount of data to be transmitted can be reduced. Alternatively, mode matrices 0, 1, 2, 3, and 4 may be group encoded and then transmitted.

The decoder decodes the converted mode matrices 1, 2, 3, and 4 shown in FIGS. 14A-14D to generate mode matrices 0, 1, 2, 3, and 4 shown in FIGS. 13A-13E. The decoder also decodes the inter-prediction mode table shown in FIG. 12 based on the decoded mode matrices 0, 1, 2, 3, and 4, and decodes the original Cb and Cr values from the conversion values based on the decoded inter-prediction mode table.

FIG. 15 is a flowchart of a method for generating transformed matrices of modes 1, 2, 3, and 4 shown in FIGS. 14A-14D and encoding information of mutual prediction modes according to an exemplary embodiment of the present invention.

In step S1510, the mode 0 matrix is multicode encoded.

In step S1520, the values '0' corresponding to the macroblocks having the values '1' in the mode 0 matrix are removed from the mode matrices 1, 2, 3, and 4 shown in FIGS. 13B-13E, and the first transformed mode matrices are formed. The first transformed matrix of mode 1, i.e. the transformed mode 1 matrix shown in FIG. 14A undergoes group coding.

In step S1530, the values '0' corresponding to the macroblocks having the values '1' in the mode 1 matrix are removed from the first transformed mode matrices 2, 3 and 4, and the second transformed mode matrices are formed. The second transformed matrix of mode 2, i.e. the transformed mode 2 matrix shown in FIG. 14B is group encoded.

In step S1540, the values '0' corresponding to the macroblocks having the values '1' in the mode 2 matrix are removed from the second transformed mode matrices 3 and 4, and the third transformed mode matrices are formed. The third transformed matrix of mode 3, i.e. the transformed mode 3 matrix shown in FIG. 14C is group encoded.

In step S1550, the values '1' in the mode 3 matrix are deleted from the third transformed mode 4 matrix, and a fourth transformed mode matrix is formed. The fourth transformed matrix of mode 4, i.e. the transformed mode 3 matrix shown in FIG. 14D is group encoded. The values in the last mode matrix are '1'. Thus, although the transformed matrix of mode 4 does not include information, the original matrix of modes can be restored using the information of the transformed matrix of modes. Therefore, the transformed mode 4 matrix may not be additionally encoded. Alternatively, step S1550 may be skipped.

Alternatively, the transformed mode matrices may be formed in an order different from the order in the present exemplary embodiment.

After steps S1510, S1520, S1540, and S1550 are completed, the multi-coding cross-prediction information is inserted into the image header of the bit stream and transmitted.

Mutual prediction using correlation between the components of the color signal of the color image, i.e. between Cb and Cr is described in the present exemplary embodiment. However, the present invention can be applied between two arbitrary regions in any color space in order to increase compression efficiency. For example, the present invention can be applied to inter-prediction using correlation between regions in a different color space, i.e. between Y and Cb or Y and Cr in the YCbCr color space.

Next, a decoder according to an exemplary embodiment is described with reference to FIG.

When the encoded bitstream is input to the decoder, the bitstream is entropically encoded, reconfigured, converted back, and input to the inverse transformer 708 of the color signal components. If the input is a component of the luminance signal, the input is ignored. If the input is a component of the color signal, the input is input to the inverter 710 of the components of the color signal.

The mutual prediction mode determination unit (not shown) restores the group matrices of the mode matrices extracted from the input bitstream image header in order from the mode 0 matrix, generates the mutual prediction mode table in the image group segment shown in Fig. 12, determines the mutual prediction mode applied to each macroblock, based on a table of mutual forecasting modes, and introduces a specific mutual forecasting mode in the inverse transform Tel 710 chrominance components.

The inverse transformer 710 of the chroma signal components generates the chroma signal components Cb and Cr from the decoded transform values using a certain mutual prediction mode.

16 is a flowchart of a decoding method according to an exemplary embodiment of the present invention.

In step S1610, entropic decoding is performed. In step S1620, inverse quantization is performed. At step S1630, the inverse transformation is performed. After that, the original mode matrices are restored from the transformed mode matrices shown in Fig. 13. A cross-prediction mode table indicating a cross-prediction mode applied in each macroblock in a predetermined group segment, for example, an image group segment.

In step S1640, the inter-prediction mode applied to the corresponding macroblock is determined from the generated inter-prediction mode table, and the decoded transform values are inversely converted according to the determined inter-prediction mode to calculate the color components Cb and Cr. In step S1650, motion compensation is performed to perform decoding. In the case of intra prediction, step S1650 is omitted.

As described above, in the method and apparatus for encoding and decoding color images using the correlation between the components of the color signal according to the present invention, a correlation between the components of the color signal of the movie image can be found to remove unnecessary signal components. Thus, the compression efficiency of motion pictures can be improved. In order to remove unnecessary components by finding a correlation between the components of the chroma signal, the information of the coefficients constituting the combination of the components of the chroma signal Cb and Cr can be group encoded. Thus, the number of bits required for encoding can be significantly reduced. In addition, the color components Cb and Cr can be converted in each mutual conversion mode. Information about the mode adapted for conversion can be divided into a matrix of modes. Mode matrices may be group encoded. As a result, further efficient group coding can be achieved.

The above encoding and decoding method may be recorded as a computer program. Codes and code segments of a computer program can be easily obtained by programmers who are specialists in the field of technology to which the present invention relates. A computer program may be stored in computer-readable media and read and executed by a computer in order to perform an encoding and decoding method. Examples of computer-readable media include magnetic recording media, optical recording media, and wave carrier.

Although the present invention is specifically shown and described with reference to its exemplary embodiments, those skilled in the art should understand that various changes in form and content can be made without departing from the spirit and scope of the present invention as defined by the appended claims inventions.

Industrial applicability

The devices and methods in accordance with the present invention can be applied to the encoding and decoding of color image data having the YCbCr format into a smaller amount of data by finding a correlation between the color components Cb and Cr of the color image data.

Claims (90)

1. A color image encoding device, comprising: a color signal component converter that generates a plurality of conversion values by multiplying the color video signal components Cb and Cr of the color video by predetermined coefficients and summing the multiplication results, selects two conversion values having the lowest costs calculated by preliminarily a certain cost function, and displays the two conversion values that are selected; and an entropic encoder performing entropic encoding of two conversion values, in which a predetermined cost function is one of the cost functions that determines the distortion depending on the transmission speed, the sum function of the absolute difference values, the sum function of the absolute transformed difference, the sum function of the difference square and the mean function absolute difference. 2. The encoding device according to claim 1, in which the components of the chrominance signal Cb and Cr are converted and quantized components of the chrominance signal. 3. The encoding device according to claim 1, further comprising a converter that converts the conversion values that are output from the converter of the components of the color signal; and a quantizer that quantizes transform values that are derived from the converter. 4. The encoding device according to claim 1, in which the Converter of the components of the chroma signal generates conversion values using the following equation: conversion value = a × Cb + b × Cr + c, where a, b, and c are predefined coefficients, and the set of sets (a, b, c) are predefined. 5. The encoding device according to claim 1, in which the Converter of the components of the chroma signal contains: a transmitter of conversion values, which generates the conversion values by multiplying the components of the chroma signal Cb and Cr by coefficients from a plurality of sets of coefficients and summing the multiplication results; a cost calculator that calculates the costs of the conversion values using a predefined cost function; and a determinant that selects and outputs the two conversion values having the least cost. 6. The device according to claim 1, in which the Converter of the components of the chroma signal is subjected to group coding information on the sets of predefined coefficients corresponding to the two conversion values that have the lowest cost. 7. A method for encoding a color image, comprising the steps of: generating a plurality of conversion values by multiplying the color signal components Cb and Cr of the color video by predetermined coefficients and summing the multiplication results; selecting two conversion values having the lowest costs calculated by a predetermined cost function; and perform entropic coding of the two transformation values that are selected, in which the predefined cost function is one of the cost functions that determines the distortion depending on the transmission speed, the sum function of the absolute difference values, the sum function of the absolute transformed difference, the sum function of the difference square and the mean function absolute difference. 8. The encoding method according to claim 7, in which the components of the chrominance signal Cb and Cr are converted and quantized components of the chrominance signal. 9. The encoding method according to claim 7, further comprising converting and quantizing the conversion values if the components of the chrominance signal Cb and Cr are not transformed and not quantized components of the chrominance signal. 10. The encoding method according to claim 7, further comprising the step of performing motion prediction and motion compensation for mutual prediction. 11. The encoding method according to claim 7, in which the conversion values are generated using the following equation: conversion value = a × Cb + b × Cr + s, where a, b, and c are predefined coefficients, and the set of sets (a, b, c) are predefined by the user. 12. The encoding method according to claim 7, in which the predefined cost function is one of the cost functions that determines the distortion depending on the transmission speed, the sum function of the absolute difference values, the sum function of the absolute transformed difference, the sum function of the difference square and the mean absolute difference function . 13. The encoding method according to claim 7, in which the formation of a plurality of conversion values comprises the steps of: multiplying the components of the chrominance signal Cb and Cr by coefficients from a plurality of sets of coefficients and summing the multiplication results with a coefficient c; and calculating the costs of the conversion values using a predetermined cost function, the coefficients a and b from the set of coefficient sets (a, b, c); and summarize the results of the multiplication with coefficient c. 14. The encoding method according to claim 1, in which information about the sets of coefficients corresponding to the two conversion values that have the lowest cost, is subjected to group coding. 15. A device for decoding a color image, comprising: an entropic decoder that entropically decodes an encoded bitstream and outputs decoded data; and an inverter of components of the color signal, which ignores the decoded data if the decoded data is a component of the luminance signal, and extracts information about the coefficients that are multiplied and added to the components of the color signal Cb and Cr to form and output the components of the color signal Cb and Cr, if the decoded data is the components of the chroma signal, in which the inverter of the components of the chroma signal extracts information indicating which set of factors is used to encode the components of the chrominance signal Cb and Cr, to form the components of the chrominance signal Cb and Cr, and the information is group encoded and transmitted. 16. The decoding device according to clause 15, further comprising: an inverse quantizer that inverse quantizes the decoded data; and an inverse converter that inversely converts the decoded data. 17. A method for decoding a color image, comprising stages in which: entropically decode the encoded bit stream to generate decoded data; ignore decoded data if the decoded data is a component of the luminance signal; and extracting information about the coefficients that are multiplied and summed with the components of the color signal Cb and Cr to generate and output the components of the color signal Cb and Cr, if the decoded data is the components of the color signal, in which the inverse component converter of the color signal extracts information indicating what set of coefficients is used to encode the components of the chrominance signal Cb and Cr, to form the components of the chrominance signal Cb and Cr, moreover, the information Gaeta-length coded and transmitted. 18. The decoding method according to claim 17, further comprising the step of decoding the decoded data back and quantizing. 19. The decoding method according to claim 17, further comprising the step of: performing motion compensation for mutual prediction. 20. An encoding device for encoding a color image, wherein the encoding device comprises: a color signal component converter, which converts the color signal component of the color image in each of at least two mutual prediction modes, calculates costs for the conversion values in each of at least of at least two mutual forecasting modes using a predefined cost function, selects one of at least two mutual forecasting modes and based on the costs and outputs of the transformation inter-prediction mode, which is selected; and an entropic encoder that performs entropic encoding of the conversion values that are output by the color component component converter, in which the color component component converter selects one of at least two mutual prediction modes in a group segment of a predetermined macroblock, and information on the mutual prediction mode, selected for a predefined macroblock, is encoded in a group segment of a predefined group containing th many blocks. 21. The coding apparatus of claim 20, wherein the mutual prediction mode information for a plurality of blocks of a predetermined group is classified into a plurality of mode matrices, and a plurality of mode matrices are encoded. 22. The coding apparatus of claim 21, wherein the plurality of mode matrices contain information about whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks. 23. The encoding device according to item 21, in which a predefined mode matrix is obtained by assigning the mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is applied, the value '1', and assigning the mode information corresponding to the block to which does not apply the mutual prediction mode corresponding to the current mode matrix, the value 'O'. 24. The coding apparatus of claim 20, wherein the mutual prediction mode information for a plurality of blocks of a predetermined group is classified into a plurality of mode matrices in each mode, the plurality of mode matrices are placed in a specific order, information about the next mode matrix is converted based on the mode information of the previous mode matrix, and the transformed mode matrix is encoded. 25. The coding apparatus of claim 24, wherein the plurality of mode matrices contain information about whether the cross-prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks, and the next mode matrix is converted by deleting information about the block to which it is applied the mutual prediction mode of the previous mode matrix, from information about the next mode matrix based on information about the previous mode matrix. 26. The encoding device according to claim 25, wherein the predetermined mode matrix is obtained by assigning the mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning the mode information corresponding to the block to which does not apply the mutual prediction mode corresponding to the current mode matrix, the value is '0'. 27. The encoding device according to p, in which the information about the block to which the mutual prediction mode of the previous mode matrix is applied is carried out by assigning the information about the block to the value '0'. 28. The encoding device according to claim 20, wherein the plurality of blocks are macroblocks, and the predefined group is an image. 29. The encoding device according to claim 20, in which the components of the color signal are Cb and Cr, which are pre-converted and pre-quantized. 30. The encoding device according to claim 20, further comprising: a converter that converts the conversion values output from the converter of the components of the color signal; and a quantizer that quantizes transform values output from the converter. 31. The encoding device according to claim 20, in which the predefined cost function is one of the cost functions that determines the distortion depending on the transmission speed, the sum function of the absolute difference values, the sum function of the absolute transformed difference, the sum function of the difference square and the mean absolute difference function . 32. The encoding device according to claim 20, in which the Converter of the components of the color signal contains: a table of mutual prediction modes, which stores a table of mutual prediction modes containing at least two mutual prediction modes; a conversion value calculator that calculates conversion values of the color signal components Cb and Cr of the color image in each of the at least two modes based on a table of mutual prediction modes; and a mode selector that selects a mutual prediction mode in which the conversion values have the lowest costs calculated by a predetermined cost function. 33. The encoding device according to claim 20, further comprising: a group encoding encoder that performs group encoding of the mutual prediction mode information that is selected. 34. An encoding method for encoding a color image, the encoding method comprising the steps of: converting the color signal components of the color image in each of at least two mutual prediction modes; calculating costs for the conversion values in each of the at least two mutual prediction modes using a predetermined cost function; selecting one of at least two cost based mutual prediction modes; and perform entropic coding of the output values of the conversion of the mutual prediction mode that is selected, in which the selection of one of the at least two mutual prediction modes is performed in a group segment of a predefined macroblock, and information about the mutual prediction mode selected for the predefined macroblock is encoded in a segment of a predefined group containing many blocks. 35. The encoding method according to clause 34, in which information about the mutual prediction mode for many blocks of a predefined group is classified into many mode matrices, and many mode matrices are encoded. 36. The encoding method according to clause 35, in which the matrix mode contain information about whether the mutual prediction mode corresponding to the current matrix mode is applied to each of the many blocks. 37. The encoding method according to claim 35, wherein a predetermined mode matrix is obtained by assigning mode information corresponding to a block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning mode information corresponding to the block to which does not apply the mutual prediction mode corresponding to the current mode matrix, the value is '0'. 38. The coding method according to clause 34, in which the mutual prediction mode information for a plurality of blocks of a predetermined group is classified into a plurality of mode matrices in each mode, a plurality of mode matrices are placed in a specific order, information about the next mode matrix is converted based on the previous mode information mode matrix, and the transformed mode matrix is encoded. 39. The encoding method according to claim 38, wherein the plurality of mode matrices contain information about whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks, and the next mode matrix is transformed by deleting information about the block to which it is applied the mutual prediction mode of the previous mode matrix, from information about the next mode matrix based on information about the previous mode matrix. 40. The encoding method according to claim 39, wherein the predetermined mode matrix is obtained by assigning the mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is applied, the value '1', and assigning the mode information corresponding to the block to which does not apply the mutual prediction mode corresponding to the current mode matrix, the value is '0'. 41. The encoding method according to claim 40, wherein deleting information about the block to which the mutual prediction mode is applied to the previous mode matrix is carried out by assigning the information about the block to the value '0'. 42. The encoding method according to claim 34, wherein the plurality of blocks are macroblocks, and the predefined group is an image. 43. The encoding method according to clause 34, in which the components of the color signal are Cb and Cr, which are pre-converted and pre-quantized. 44. The encoding method according to clause 34, further comprising converting and quantizing the conversion values of the mutual prediction mode, which is selected if the components of the chrominance signal Cb and Cr contain values that are not converted and not quantized. 45. The decoding method according to clause 34, further comprising the step of performing mutual prediction. 46. The encoding method according to clause 34, in which the predefined cost function is one of the cost functions that determines the distortion depending on the transmission speed, the sum function of the absolute difference values, the sum function of the absolute transformed difference, the sum function of the difference square and the mean absolute difference function . 47. The encoding method according to clause 34, further comprising the step of calculating the conversion values of the color components Cb and Cr of the color image in each mode, the selection of one of the at least two mutual prediction modes comprises selecting a mode mutual forecasting, in which the conversion values have the lowest costs, calculated by means of a predefined cost function. 48. The encoding method according to clause 34, further comprising the step of performing group encoding of the information about the mutual prediction mode that is selected. 49. A decoding device for decoding an encoded color image, wherein the decoding device comprises: an entropic decoder that entropically decodes an input bitstream; and an inverse component of the color signal, which restores the original components of the color signal based on the information of the mutual prediction mode applied to the current block having a predetermined size, and the information of the mutual prediction mode is extracted from the input bit stream, while the information of the mutual prediction mode indicates the mutual mode forecasting, which is selected from at least two modes of mutual forecasting, and applies to the current block, and the original components of the color signal are obtained from the conversion values corresponding to the mutual prediction mode applied to the current block, in which the information of the mutual prediction mode for the input bit stream is a set of matrixes of modes into which the information of the mutual prediction mode for multiple blocks is classified into each mode. 50. The decoding apparatus of claim 49, wherein the plurality of mode matrices comprise information on whether a mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks. 51. The decoding apparatus of claim 49, wherein a predetermined mode matrix is obtained by assigning mode information corresponding to a block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning mode information corresponding to the block to which does not apply the mutual prediction mode corresponding to the current mode matrix, the value is '0'. 52. The decoding apparatus of claim 49, wherein the mutual prediction mode information extracted from the input bitstream is a transformed mode matrix generated by classifying the mutual prediction mode information for a plurality of blocks of a predetermined group into a plurality of mode matrices in each mode, placement multiple mode matrices in a specific order and converting information about the next mode matrix based on the mode information for the previous matrix Ima. 53. The decoding apparatus according to claim 52, wherein the plurality of mode matrices contain information about whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks, and the next mode matrix is converted by deleting information about the block to which it is applied the mutual prediction mode of the previous mode matrix, from information about the next mode matrix based on information about the previous mode matrix, wherein the decoding device decodes the transformed matrices s modes in a specific order to restore the original matrix of modes. 54. The decoding apparatus of claim 52, wherein the predetermined mode matrix is obtained by assigning mode information corresponding to a block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning mode information corresponding to the block to which does not apply the mutual prediction mode corresponding to the current mode matrix, the value is '0'. 55. The encoding device according to item 54, in which the information about the block to which the mutual prediction mode of the previous mode matrix is applied is deleted by assigning the information about the block to the value '0'. 56. The decoding apparatus of claim 49, wherein the current block having a predetermined size is a macroblock, and the mode matrices comprise mutual prediction mode information for the macroblock of an image group segment. 57. The decoding apparatus of claim 49, wherein the components of the chrominance signal are Cb and Cr. 58. The decoding apparatus of claim 49, further comprising: an inverse quantizer that inversely quantizes decoded data; and an inverse converter that inversely converts the decoded data. 59. The decoding apparatus of claim 49, wherein the decoding apparatus extracts the mutual prediction mode information encoded and transmitted according to the group encoding method to calculate the color signal components based on the mutual prediction mode information. 60. A decoding method for decoding an encoded color image, the decoding method comprising the steps of: performing entropic encoding of an input bit stream; and restoring the original components of the color signal based on the information of the mutual prediction mode applied to the current block having a predetermined size, the information of the mutual prediction mode being extracted from the input bit stream, while the information of the mutual prediction mode indicates the mutual prediction mode, which is selected from at least two modes of mutual forecasting, and is applied to the current block, and the initial components of the color signal I get Xia from conversion values corresponding to the applied current block, inter-prediction mode, wherein the inter-prediction mode information extracted from the input bitstream is a plurality of mode planes into which inter-prediction mode information for the plurality of blocks is classified in each mode. 61. The decoding method of claim 60, wherein the plurality of mode matrices comprise information on whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks. 62. The decoding method of claim 60, wherein the predetermined mode matrix is obtained by assigning mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning mode information corresponding to the block to which does not apply the mutual prediction mode corresponding to the current mode matrix, the value is '0'. 63. The decoding method of claim 60, wherein the mutual prediction mode information extracted from the input bitstream is a transformed mode matrix generated by classifying the mutual prediction mode information for a plurality of blocks of a predetermined group into a plurality of mode matrices in each mode, placement multiple mode matrices in a specific order and converting information about the next mode matrix based on mode information about the previous mode matrix. 64. The decoding method of claim 63, wherein the plurality of mode matrices contain information about whether the cross-prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks, and the next mode matrix is converted by deleting information about the block to which it is applied the mutual prediction mode of the previous mode matrix, from the information about the next mode matrix based on the information about the previous mode matrix, while the transformed mode matrices are decoded at a certain time core to restore the original matrix modes. 65. The decoding method of claim 64, wherein the predetermined mode matrix is obtained by assigning the mode information corresponding to the block to which the mutual prediction mode corresponding to the current mode matrix is applied, the value '1', and assigning the mode information corresponding to the block to which does not apply the mutual prediction mode corresponding to the current mode matrix, the value is '0'. 66. The decoding method according to claim 65, wherein deleting information about the block to which the mutual prediction mode of the previous mode matrix is applied is carried out by assigning the information about the block to the value '0'. 67. The decoding method according to claim 60, wherein the current block having a predetermined size is a macroblock, and the mode matrices contain mutual prediction mode information for the macroblock of an image group segment. 68. The decoding method of claim 60, wherein the components of the chrominance signal are Cb and Cr. 69. The decoding method of claim 60, further comprising the step of decoding the decoded data back and quantizing. 70. The decoding method of claim 60, further comprising the step of performing mutual prediction. 71. The decoding method of claim 60, wherein the mutual prediction mode information encoded and transmitted according to the group coding method is extracted to calculate the color signal components based on the mutual prediction mode information. 72. A computer-readable recording medium having a computer program embodied therein for a computing device to execute a color image encoding method, the encoding method comprising the steps of: generating a plurality of conversion values by multiplying the color signal components Cb and Cr of the color video by predetermined coefficients and summation of multiplication results; selecting two conversion values having the lowest costs calculated by a predetermined cost function; and perform entropic coding of the two transformation values that are selected, in which the predefined cost function is one of the cost functions that determines the distortion depending on the transmission speed, the sum function of the absolute difference values, the sum function of the absolute transformed difference, the sum function of the difference square and the mean function absolute difference. 73. A computer-readable recording medium having a computer program embodied therein for executing by a computing device a method for encoding a color image, the encoding method comprising the steps of: converting the color signal components of a color image in each of at least two mutual prediction modes; calculating costs for the conversion values in each of the at least two mutual prediction modes using a predetermined cost function; selecting one of at least two cost based mutual prediction modes; and perform entropic coding of the output values of the conversion of the mutual prediction mode that is selected, in which the selection of one of the at least two mutual prediction modes is performed in a group segment of a predefined macroblock, and information about the mutual prediction mode selected for the predefined macroblock is encoded in a segment of a predefined group containing many blocks. 74. The computer-readable recording medium of claim 73, wherein the mutual prediction mode information for a plurality of blocks of a predetermined group is classified into a plurality of mode matrices, and a plurality of mode matrices are encoded. 75. The computer-readable recording medium according to claim 74, wherein the mode matrices contain information about whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks. 76. The computer-readable recording medium according to claim 74, wherein the predetermined mode matrix is obtained by assigning mode information corresponding to a block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning mode information corresponding to the block, to which the mutual prediction mode corresponding to the current mode matrix is not applied, the value is '0'. 77. The computer-readable recording medium according to claim 73, wherein the mutual prediction mode information for a plurality of blocks of a predetermined group is classified into a plurality of mode matrices in each mode, a plurality of mode matrices are placed in a specific order, information about the next mode matrix is converted based on the mode information about previous mode matrix, and the transformed mode matrix is encoded. 78. The computer-readable recording medium of claim 77, wherein the plurality of mode matrices contain information about whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks, and the next mode matrix is converted by deleting information about the block to which the mutual prediction mode of the previous mode matrix is applied, from the information about the next mode matrix based on the information about the previous mode matrix. 79. The computer-readable recording medium of claim 78, wherein the predetermined mode matrix is obtained by assigning mode information corresponding to a block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning mode information corresponding to the block, to which the mutual prediction mode corresponding to the current mode matrix is not applied, the value is '0'. 80. The computer-readable recording medium of claim 79, wherein deleting the block information to which the mutual prediction mode is applied to the previous mode matrix is performed by assigning the block information to '0'. 81. A computer-readable recording medium having a computer program embodied therein for executing by a computing device a method for decoding a color image, the decoding method comprising the steps of: decoding an encoded bit stream entropically to generate decoded data; ignore decoded data if the decoded data is a component of the luminance signal; and extracting information about the coefficients, which are multiplied and summed with the components of the color signal Cb and Cr, to generate and output the components of the color signal Cb and Cr, if the decoded data is the components of the color signal, in which the information indicates which set of coefficients is used for to encode the components of the chrominance signal Cb and Cr, to form the components of the chrominance signal Cb and Cr, and the information is group encoded and transmitted. 82. A computer-readable recording medium having a computer program embodied therein for executing by a computing device a method for decoding a color image, the decoding method comprising the steps of: entropically encoding an input bit stream; and restoring the original components of the color signal based on the information of the mutual prediction mode applied to the current block having a predetermined size, the information of the mutual prediction mode being extracted from the input bit stream, while the information of the mutual prediction mode indicates the mutual prediction mode, which is selected from of at least two mutual prediction modes, and is applied to the current block, and the initial components of the color signal are obtained are derived from the conversion values corresponding to the mutual prediction mode applied to the current block, in which the information of the mutual prediction mode extracted from the input bitstream is a set of mode matrices into which the information of the mutual prediction mode for multiple blocks is classified in each mode. 83. The computer-readable recording medium of claim 82, wherein the plurality of mode matrices contain information about whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks. 84. The computer-readable recording medium of claim 82, wherein the predetermined mode matrix is obtained by assigning mode information corresponding to a block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning mode information corresponding to the block, to which the mutual prediction mode corresponding to the current mode matrix is not applied, the value is '0'. 85. The computer readable recording medium of claim 82, wherein the mutual prediction mode information extracted from the input bitstream is a transformed mode matrix generated by classifying the mutual prediction mode information for a plurality of blocks of a predetermined group into a plurality of mode matrices in each mode, placing multiple mode matrices in a specific order and converting information about the next mode matrix based on mode information about the previous matrices mode. 86. The computer-readable recording medium of claim 85, wherein the plurality of mode matrices contain information about whether the mutual prediction mode corresponding to the current mode matrix is applied to each of the plurality of blocks, and the next mode matrix is converted by deleting information about the block to which the mutual prediction mode of the previous mode matrix is applied, from the information about the next mode matrix based on the information about the previous mode matrix, while the transformed mode matrices are decoded in In order to restore the original mode matrices. 87. The computer-readable recording medium of claim 86, wherein a predetermined mode matrix is obtained by assigning mode information corresponding to a block to which the mutual prediction mode corresponding to the current mode matrix is applied, a value of '1', and assigning mode information corresponding to the block, to which the mutual prediction mode corresponding to the current mode matrix is not applied, the value is '0'. 88. The computer-readable recording medium of claim 87, wherein the block information to which the mutual prediction mode of the previous mode matrix is applied is deleted by assigning the block information to '0'. Priority on points: December 30, 2004 according to claims 1-19, 72, 81; 04/02/2005 according to claims 20-71, 73-80, 82-88.

Images