JPH1013859A - High efficiency encoder for picture, high efficiency decoder for picture and high efficiency encoding and decoding system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain high efficiency encoder reducing a chrominance noise spreading range and reducing chrominance distortion by separating a luminance component and a color difference component from each other, then blocking according to a prescribed sampling density ratio and respectively encoding them. SOLUTION: A luminance and color difference components separating blocking part 1 separates the luminance component and the color difference component of a digitally inputted picture signal 100 from each other and then blocks them according to the prescribed sampling density ratio. The respective blocked pictures 101 and 102 are orthogonally transformation- encode-processed to a frequency area respectively by a luminance component transforming part 3 and a color difference component transforming part 4, and their transforming coefficients are quantized respectively by quantization parts 5a and 5b. In addition respective obtained quantization coefficients 107 and 108 are outputted as respective variable length encoded words 109 and 110 by variable length encoding parts 6a and 6b. Continually a variable length encoded word 111 added by an adder 9 is inputted to a buffer 11 to be quantize- controlled corresponding to the bit quantity of a buffer output 113 and multiplexed by a multiplexing part 12 in addition to be transmitted. Thereby, it is possible to execute high efficiency coding which reduces chrominance noise and chrominance distortion.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image encoding system which performs high-efficiency encoding or decoding of an image, such as an image broadcasting system or a video disk system, and which can be provided to a system for efficiently transmitting or storing the image. And a decoder.

[0002]

2. Description of the Related Art As a conventional representative high-efficiency encoding system, there is MPEG2 which is an international standard system studied in ISO / IEC JTC1 / SC29 / WG11. For example, in the April 1995 issue of "The Journal of the Institute of Television Engineers, Image Information Engineering and Broadcasting Technology", MPEG is described as a special theme. See p. From 29-60, "3
MPEG-2 encoding method is introduced as "-2 video compression". Hereinafter, the conventional high-efficiency coding method will be described based on the above description.

FIG. 9 is mainly used in MPEG2 4:
It is an explanatory view of an image format called 2: 0, and shows a luminance and color difference format. In MPEG2, these formats are not dynamically changed, and encoding or decoding is performed in a state where the format is fixed to one of the formats. At 4: 2: 0, the sample density of the chrominance component is half of the sample density of the luminance component in both the horizontal and vertical directions, which is extremely low. This is because an attempt is made to obtain an information compression effect by utilizing the fact that the resolution discriminating ability of a person is higher than the luminance.

FIG. 10 shows the basic configuration of the MPEG encoder described in the above description, and FIG. 11 also shows the basic configuration of the MPEG decoder described in the above description. In the figure, 50 is an A / D converter, 52 is a DCT, 54 is a quantizer, 57
Is a variable length coding unit, 58 is a transmission buffer, 53 is a rate control unit, 62 is a reception buffer, 63 is a variable length decoding unit, 5
6 is an inverse DCT unit, 66 is a subtractor, 51 is an inter (intra-frame) / intra (inter-frame) switching selector, 5
9 is an adder, 60 is a frame memory, 61 is a motion compensation prediction unit for performing motion compensation vector estimation and motion compensation, 67
Is a motion compensation prediction unit. Also, 200 is digitized image data, 204 is a transform coefficient by DCT, 2
06 is a quantization index of a transform coefficient, 215 is a coded bit stream, 207 is a signal indicating the amount of information generation, 2
06 is the quantization index of the transform coefficient subjected to the variable length decoding, 208 is the inversely quantized transform coefficient, 202 is the prediction error image data, 209 is the image data returned to the pixel space area by the inverse DCT, 213 is the prediction Image data, 21
0 is the decoded image data.

The operation of the encoder will be described with reference to FIG. The input image signal 200 is digitized in the A / D converter 50. The input screen is encoded by motion compensation prediction + DCT encoding. Input image data 201
And the motion compensation predicted image data 213 generated by motion prediction from the reference screen,
02 is obtained. The prediction error signal is converted into a transform coefficient 204 in the spatial frequency domain by the DCT unit 52 in block units of 8 pixels × 8 lines, and quantization is performed in the quantization unit 54. In the case of intra coding without performing motion compensation prediction, the input image data 201 is directly subjected to DCT coding. This switching is performed by the selector 51. The quantization coefficient 20 is used later as a reference screen for motion compensation prediction.
6 is subjected to inverse quantization (55) and inverse DCT (56), and is added to the motion compensated prediction signal 213 (59), that is, an image is decoded by local decoding, and is stored in the frame memory 60. The quantized 8 × 8 DCT coefficients are sequentially scanned from low frequency components to become one-dimensional information, and then the motion vector 1
Variable length coding is performed together with other coding information such as H.24 (57). To keep the variable code generation constant,
The generated code amount 207 is grasped by monitoring the output buffer 58, and quantization control is performed by feedback (5).
3) The method is general. The output of buffer 58 is encoded bitstream 216.

The operation of the decoder will be described with reference to FIG. The decoding process is basically the reverse operation of the encoder.
First, the encoded bit stream 216 is stored in the buffer 6
2 is stored. The data in the buffer 62 is read and variable length decoding is performed (63). In this process, DC
The information 208 of the T coefficient, the motion vector 124, etc.
Separated. The decoded 8 × 8 quantized DCT coefficient 2
06 is restored to the DCT coefficient 208 by the inverse quantization 55, and is converted to the pixel space data 209 by the inverse DCT 56.
At the time of intra coding, a decoded image has been obtained at this stage. When the motion compensation prediction is being performed, the image is decoded by adding the motion compensation prediction image data 213 generated by the motion compensation prediction from the reference screen. The decoded image is stored in the frame memory 60 for use as a reference screen in a subsequent decoding process as needed.

[0007]

The conventional typical encoder is configured as described above, and the encoding of the input image is basically performed in block units of DCT, and the sample density ratio of the luminance / chrominance components is used. Is statically fixed at 4: 2: 0.
Therefore, there is a problem that the deterioration of the image quality due to the compression spreads in block units. This is because noise generated in a specific transform coefficient due to quantization spreads to the entire block by inverse DCT. Furthermore, since the sample density of the color difference component is generally lower than the sample density of the luminance component, there is also a problem that this deterioration is noticeably observed in the color component. Increasing the sample density of the color difference component mitigates the phenomenon that color noise is particularly noticeable, but has a dilemma of increasing the number of samples to be coded and disadvantageous in increasing the compression efficiency.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and it is an object of the present invention to reduce a color noise which becomes remarkable when a compression ratio is increased, and to obtain a high-efficiency code capable of obtaining a higher-quality coded image. The purpose is to obtain a decoder and a decoder.

[0009]

A high-efficiency encoder for an image according to the present invention encodes a target image signal and outputs an encoded bit stream. A luminance / chrominance component separation unit that separates the luminance signal into color difference signals and blocks them at a predetermined sample density ratio; a luminance component conversion unit that orthogonally transforms the separated luminance signal into a frequency domain; A luminance signal quantizer for quantizing the signal, a luminance signal encoder for encoding the quantized luminance signal, a color difference component converter for orthogonally transforming the separated color difference signal into the frequency domain, A chrominance signal quantization unit that quantizes the converted chrominance signal, a chrominance signal encoding unit that encodes the quantized chrominance signal, and a multiplex that multiplexes a luminance signal encoding output and a chrominance signal encoding output. Conversion The multiplexed output was coded bit stream comprises a.

[0010] Further, a luminance signal local decoding means for inversely quantizing the quantized luminance signal, further inversely transforming the quantized luminance signal and connecting it back to the separated luminance signal via a luminance predictor, A chrominance signal local decoding means for performing inverse quantization, further inverse transform, and feeding back to the separated chrominance signal via a chrominance predictor is added.

Further, a luminance inverse quantization / inverse transform unit for inversely quantizing the quantized luminance signal and further performing inverse transform, and a color difference inverse quantization for inversely quantizing the quantized color difference signal and further performing inverse transform. An inverse transform unit, a luminance / chrominance component combination block unit that combines a luminance signal after inverse transformation and a chrominance signal after inverse transformation into a block signal, and a luminance / chrominance signal after integration through a predictor A local decoding means for subtractive feedback connection is added to the target image signal.

Still further, the predictor in the local decoding means includes a motion compensation prediction unit. The motion compensation prediction unit separates the prediction into the luminance component motion prediction and the color difference component motion prediction, An error from the color difference signal is calculated, and the calculated prediction error is weighted and added, and a subtraction feedback connection is made to the target image signal.

The high-efficiency image decoder according to the present invention comprises:
In a picture decoder for obtaining an image signal from a coded bit stream to be decoded, a demultiplexer for separating an input coded bit stream into coded data of a luminance signal and a chrominance signal; Luminance decoding unit for decoding the decoded data, a luminance inverse quantization unit for inversely quantizing the decoded luminance decoded signal, and a luminance inverse transform unit for inverse orthogonal transforming the inversely quantized luminance decoded signal A color difference decoding unit that decodes the separated color difference encoded data, a color difference inverse quantization unit that inversely quantizes the decoded color difference decoded signal, and an inverse quantization unit that inversely quantizes the inversely quantized color difference decoded signal. A chrominance inverse transform unit for performing orthogonal transform, and a luminance / integration unit that combines the luminance inverse orthogonal transformed luminance signal output and the chrominance inverse orthogonal transformed chrominance signal output.
The luminance / color difference combined output provided with a color difference component combining section was used as a reproduced image signal.

Further, a luminance predicting means for predicting the luminance of the luminance signal subjected to the luminance inverse orthogonal transformation via a luminance adder and connecting the predicted luminance signal back to the luminance adder,
A chrominance prediction means for predicting chrominance of the chrominance signal subjected to the chrominance inverse orthogonal transform via a chrominance adder and connecting the chrominance signal after the prediction to the chrominance adder is added.

Further, an output of a luminance / chrominance combining unit for combining the luminance signal subjected to the luminance inverse orthogonal transform and the color difference signal subjected to the color inverse inverse orthogonal transform is predicted via an image adder. Is added to the above-mentioned image adder.

Further, the luminance / chrominance combining unit combines luminance and chrominance according to a specified sample density ratio included in the coded bit stream to be decoded.

A high-efficiency encoding / decoding system for an image according to the present invention comprises a luminance / chrominance separation section for separating a target image signal into a luminance signal and a chrominance signal to form a block, and a frequency domain of the separated luminance signal. A luminance signal quantizing unit for quantizing the luminance signal after the orthogonal transformation into a luminance signal, a luminance signal encoding unit for encoding the quantized luminance signal, and an orthogonal transformation of the separated color difference signal into a frequency domain. A chrominance signal quantization unit for quantizing the subsequent chrominance signal, a chrominance signal encoding unit for encoding the quantized chrominance signal, and multiplexing for multiplexing a luminance signal encoding output and a chrominance signal encoding output. A high-efficiency encoder for an image having a unit, a multiplexing / demultiplexing unit for separating an encoded bit stream from the high-efficiency encoder for an image into encoded data of a luminance signal and a chrominance signal, and the separated luminance. Decode encoded data A luminance decoding unit that dequantizes the decoded luminance decoded signal, a chrominance decoding unit that decodes the separated chrominance-encoded data, A chrominance inverse quantization unit that inversely quantizes the chrominance decoded signal, a luminance signal output after inverse orthogonal transformation of the inversely quantized luminance decoded signal, and an inverse orthogonal transformation of the inversely quantized chrominance decoded signal after inverse orthogonal transformation. And a high-efficiency image decoder having a luminance / chrominance combining unit for combining with the chrominance signal output.

[0018]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1 FIG. INDUSTRIAL APPLICABILITY The high-efficiency image encoder and decoder of the present invention can be applied to a digital broadcasting system, a digital video disk, a mobile videophone, a PHS videophone, and the like that are performed via a satellite, a terrestrial wave, or a wired communication network.
FIG. 1 is a configuration diagram of a high-efficiency image encoder according to the present embodiment. In the figure, 1 is a luminance / chrominance component separation block forming unit, 3 is a luminance component conversion unit, 4 is a chrominance component conversion unit,
5a is a luminance signal quantization unit, 5b is a color difference signal quantization unit, 6
a is a luminance signal variable length coding unit, and 6b is a chrominance signal variable length coding unit. 9 is an adder, 10 is a quantization control unit, 11
Is a buffer, and 12 is a multiplexing unit. Also, 100 is digitized image data, 101 is YC-separated blocked image data of a luminance component, 102 is YC-separated blocked image data of a chrominance component, 105 is a conversion coefficient of a luminance component, and 106 is a chrominance Component conversion coefficient, 107 is a luminance component quantization coefficient, 108 is a chrominance component quantization coefficient, 109 is a luminance component variable length coded word, 110 is a chrominance component variable length coded word, 111 is luminance and chrominance A variable-length coded word of both components, 112 is a variable-length coded word sent from a buffer, 113 is a buffer output, 114 is an adaptive quantization value, 115 is a coded bit stream sent from the multiplexing unit 12, 219 Is the sample density ratio when the input image is divided into luminance / color difference components.

Next, the operation of the apparatus having the above configuration will be described. The digitized input image signal is YC-separated, that is, divided into a luminance component and a chrominance component in the luminance / chrominance component separation / blocking unit 1 and then blocked according to a predetermined sample density ratio 219. FIG.
It depicts pixel samples for one macroblock in 4: 2: 0 format used in video encoding systems such as G and H263. In the figure, X indicates a luminance pixel, and O indicates a chrominance pixel. According to the conventional DCT method, the luminance component has four sub-blocks (8 × 8).
Luminance pixels distributed over the area are divided into four 8 × 8 DCTs
In contrast to the conversion process, the color difference components distributed over four sub-blocks (8 × 8) are subjected to one 8 × 8 DCT conversion process.

On the other hand, according to this method, 101 is, for example, 8
Assuming that the luminance component is block image data composed of blocks of 8 lines of pixels, if the chrominance component is blocked at a sample density ratio with respect to the luminance component, the chrominance component block image data is 4 pixels × 4 pixels. Become a block of lines. If 101 is block image data of 16 pixels × 16 lines, 102 is 8 pixels ×
That is, it is block image data of eight lines of color difference components. Each of these block images 101 and 102 is
Each of the luminance component conversion unit 3 and the chrominance component conversion unit 4 performs an orthogonal transform coding process on a frequency domain. Therefore, according to the above numerical example, the luminance component conversion unit 3 performs conversion coding of 8 pixels × 8 lines, and the chrominance component conversion unit 4 performs conversion coding of 4 pixels × 4 lines. The obtained conversion coefficients are divided into a luminance signal quantization unit 5a for each of a luminance component and a chrominance component.
Each of the quantized coefficients 10 is quantized by the color difference signal quantizing unit 5b.
Reference numerals 7 and 108 further denote each variable-length encoded word 10 in the luminance signal variable-length encoding unit 6a and the chrominance signal variable-length encoding unit 6b.
9, 110 are output. The variable-length coded word 111 added by the adder 9 is input to the buffer 11, and
The quantization control is performed according to the bit amount of the buffer output 113. As a quantization control method, for example, a conventional method of increasing the quantization step size in order to suppress the amount of generated bits when the buffer output 113 is large, and decreasing the quantization step size in the opposite case, is used. Etc. may be used. The luminance signal variable-length coded word 109 and the chrominance signal variable-length coded word 110 stored in the buffer 11 are multiplexed by the multiplexing unit 12 and transmitted or stored as an output of the high-efficiency encoder.

In the above-described embodiment, the coding systems for the luminance component and the chrominance component are different. However, when the quantization unit, the inverse quantization unit, and the variable length coding unit can be used in common, A configuration in which a luminance component and a chrominance component are multiplexed and input, and quantization, dequantization, and variable-length coding are sequentially processed in the order of a luminance component blocked image and a chrominance component blocked image, The same effect as above can be obtained. Also, in the above embodiment, when the block image of the luminance component is 8 pixels × 8 lines, the block image of the chrominance component is 4 pixels × 4 in consideration of the sample density of the 4: 2: 0 format. Although the number of samples of the block image is small, for example, when the block image of the luminance component is 4 pixels × 4 lines or less, the block image of the chrominance component is the same as the block image of the luminance component. You can also

Embodiment 2 FIG. FIG. 3 is a configuration diagram of the high-efficiency image decoder according to the present embodiment. In the figure, 7a
Is a luminance inverse quantization unit, 7b is a chrominance inverse quantization unit, 8 is a chrominance component inverse transform unit, 27 is a luminance component inverse transform unit, 20 is a demultiplexing unit, 21a is a luminance variable length decoding unit, and 21b is a chrominance The variable length decoding unit 30 is a luminance / chrominance component combination blocking unit. Reference numeral 118 denotes an inversely-quantized transform coefficient of a luminance component; 119, an inverse-quantized transform coefficient of a chrominance component;
Reference numeral 0 denotes an image obtained by inversely converting the luminance component, reference numeral 121 denotes an image obtained by inversely converting the chrominance component, and reference numeral 128 denotes a block image obtained by combining the luminance and the chrominance. Other than the above, they are equivalent to the elements or signals of the numbers already described.

Next, the operation of the apparatus having the above configuration will be described. The decoder according to the second embodiment is a decoder corresponding to the encoder according to the first embodiment. First, a variable length coded word 109 of a luminance component and a variable length coded word 110 of a chrominance component are output from the demultiplexing unit 20 of the input coded bit stream 115. These are immediately changed by the luminance variable length decoding unit 21a and the chrominance variable length decoding unit 21b.
Are converted to the respective quantization coefficients, and the respective prediction error images 120 and 121 are obtained in exactly the same procedure as in the encoder. Further, the luminance component inverse transformed image 120 and the chrominance component inverse transformed image 121 are combined with the luminance / chrominance component according to the sample density ratio 219 used in the encoder, and a new decoded block image 128 is finally output. Is done.

Embodiment 3 FIG. A high-efficiency encoder having a local decoder for performing prediction control to which the present invention is applied will be described.
FIG. 4 is a configuration diagram of the high-efficiency image encoder according to the present embodiment. In the figure, 13 is a luminance component frame memory,
14 is a chrominance component frame memory, 15 is a luminance component motion compensation prediction unit, 16 is a chrominance component motion compensation prediction unit, and 27 is a luminance component inverse conversion unit. Reference numeral 103 denotes a prediction error image of block image data of a luminance component; 104, a prediction error image of block image data of a chrominance component; 116, a motion vector of a chrominance component; 117, a motion vector of a luminance component;
22 is a locally decoded image of the luminance component, 123 is a locally decoded image of the chrominance component, 124 is a reference image of the read chrominance component, 125 is a reference image of the read luminance component, 12
6 is a predicted image of a color difference component, and 127 is a predicted image of a luminance component.

Next, the operation of the apparatus having the above configuration will be described. The digitized input image signal is YC-separated, that is, divided into a luminance component and a chrominance component in a luminance / chrominance component separation / blocking unit, and then blocked.
At this time, the sample density ratio 219 of the luminance component and the chrominance component is sent to the multiplexing unit 12. Luminance component blocked image 1
01 and the chrominance component blocked image 102 are subtracted from the prediction image by the respective subtracters 2a and 2b, and the luminance prediction error image 103 and the chrominance prediction error image 1 of these differences are obtained.
In reference numeral 04, the luminance component conversion unit 3 and the chrominance component conversion unit 4 perform conversion coding processing to the frequency domain. The obtained transform coefficients are quantized by the luminance signal quantization unit 5a and the color difference signal quantization unit 5b for each of the luminance component and the chrominance component, and the quantization coefficients 107 and 108 are obtained. These are further output as variable length coded words 109 and 110 in the luminance variable length coding unit 6a and the color difference variable length coding unit 6b. The variable-length coded word 111 added by the adder 9 is input to the buffer 11, and quantization control is performed according to the bit amount of the buffer output 113. The output signal 115 is output after being multiplexed by the multiplexing unit 12. On the other hand, the luminance quantization coefficient 107 and the chrominance quantization coefficient 108 are inversely quantized by the inverse quantizers 7a and 7b for local decoding, and transform coefficients 118 and 119 are obtained. The inversely quantized transform coefficients 118 and 119 are inversely transformed by the chrominance component inverse transform unit 8 and the luminance component inverse transform unit 27, and are returned to the luminance / chrominance prediction error images 120 and 121, respectively, in the time domain. Further, the predicted images 12 are added by the adders 9a and 9b.
6 and 127, and the respective locally decoded images 122 and 123 are output. These are stored in the luminance component frame memory 13 and the chrominance component frame memory 14, respectively, and are used as reference images for motion compensation prediction. In the motion compensation prediction, respective motion vectors 116 and 117 are output using the reference images 124 and 125 in the motion compensation prediction units 15 and 16. The motion-compensated prediction, which has been conventionally used, may be a method of detecting a motion when a prediction error is minimized in a search range within a certain range, as it is.

In the above embodiment, the coding system for the luminance component and the chrominance component is different. However, if the quantization unit, the inverse quantization unit, and the variable length coding unit can be used in common, And the chrominance component are multiplexed and input, and the quantization, dequantization, and variable-length coding processes are performed on a luminance component block image,
The same effect as in the above embodiment can be obtained even if the processing is performed sequentially in the order of the color difference component blocked images. Also, in the above embodiment, when the block image of the luminance component is 8 pixels × 8 lines, the block image of the chrominance component is 4 pixels × 4 in consideration of the sample density of the 4: 2: 0 format. When the number of samples of the block image is small, for example, when the block image of the luminance component is 4 pixels × 4 lines or less, the block image of the chrominance component is the same as the block image of the degree component. Can be

Embodiment 4 A high-efficiency decoder corresponding to an encoder with local decoding means for performing prediction control will be described.
FIG. 5 is a configuration diagram of a high-efficiency image decoder according to the present embodiment. In the figure, 22 is a chrominance component motion compensator, and 23 is a luminance component motion compensator. Reference numeral 128 denotes a block image in which luminance and chrominance are combined. Elements other than the above are elements or signals equivalent to those of the numbers already described.

Next, the operation of the apparatus having the above configuration will be described. First, the coded bit stream 115 is converted into a variable-length coded word 10
9 and the variable-length coded word 110 of the color difference component are output separately. These are immediately transmitted to the variable length decoding unit 2 respectively.
In 1a and 21b, each quantized coefficient is converted to a corresponding one, and the prediction error images 120 and 121 are obtained by passing the subsequent elements in exactly the same procedure as in the encoder. Also, the motion vectors 116, 117 transmitted from the demultiplexing unit 20
, The reference images 124 and 12 from the chrominance component frame memory 14 and the luminance component frame memory 13, respectively.
5, the respective predicted images 126 and 127 are output. These are added to the predicted image 1 by the adders 9a and 9b.
20 and 121 to obtain respective decoded block images 122 and 123. Further, these signals 12
2 and 123 are sample density ratios 2 for which the luminance / color difference components are specified.
19 to form a new decoded block image 12
8 is finally output.

In the above-described embodiment, on the decoder side, the luminance and chrominance components are combined at the specified sample density ratio included in the input coded bit stream to be decoded. However, without using the sample density ratio 219 sent from the encoder, the luminance / chrominance component combination blocking unit 3 according to an arbitrary sample density ratio at the decoder side.
At 0, the luminance component block and the chrominance component block may be combined. This corresponds to, for example, a case where the luminance / chrominance component combination blocker 30 does not have a blocking function according to the sample density ratio used in the encoder.

Embodiment 5 Another high-efficiency encoder having local decoding means for performing prediction control will be described. FIG. 6 is a configuration diagram of the high-efficiency image encoder according to the present embodiment. In the figure, 24 is a frame memory, 25 is a motion compensation prediction unit, 129 is a motion prediction error image, 130 is a prediction error blocked image of a luminance component, 131 is a prediction error blocked image of a chrominance component, and 132 is a decoding 133 is a local decoded image, 134 is a reference image from the frame memory 24, 135 is a motion prediction image, and 136 is a motion vector. Elements other than the above are equivalent to the elements or signals of the numbers already described.

Next, the operation of the apparatus having the above configuration will be described. In the apparatus according to the present embodiment, first, a luminance / chrominance component separation / blocking unit 1 separates and blocks each of the motion compensation prediction error images 129 according to a predetermined sample density ratio and obtains a luminance component block image. 130,
Each of the conversion units 3 and 4 converts the color difference component blocked image 131. In the above-described blocking for each component, as described in Embodiment 1, in consideration of the sample density of the luminance component and the chrominance component, for example, in the case of the 4: 2: 0 format as shown in FIG. Is 8 pixels × 8 lines, the chrominance component blocks 4 pixels × 4 lines, and if the luminance component is 16 pixels × 16 lines, the chrominance component is 8 pixels × 8 lines or 4 pixels × 4 lines. Blocking may be performed on four lines. Subsequently, the obtained transform coefficients 105 and 106 are, as described in the first embodiment, the respective quantizers 5a and 5b, the respective inverse quantizers 7a and 7b, the chrominance component inverse converter 8, and the luminance component inverse converter. After decoding through 27, the respective prediction error images 120 and 121 are obtained. These prediction error images 120 and 121 are combined by the luminance / chrominance component combination blocking unit 30 to output a prediction error image 132. Reference numeral 132 denotes an adder 9c which further outputs the predicted image 1 output from the motion compensation prediction unit 25.
35 to obtain a locally decoded image 133.
The predicted image 135 is fed back for subtraction from the input digital image 100. Further, the motion vector 136 detected at the time of the motion compensation prediction is multiplexed with the buffer output 112 in the multiplexing unit 12 and transmitted as a coded bit stream 115 to the transmission path or stored and output. The operation of the quantization control unit 10 is the same as that of the first embodiment.

In the above embodiment, when the luminance component block image is 8 pixels × 8 lines, 4: 2:
Although the block image of the chrominance component is 4 pixels × 4 lines in consideration of the sample density of the 0 format, when the number of samples of the block image is small, for example, the block image of the luminance component is 4 pixels × 4 lines. In the following cases,
The block image of the color difference component may be the same as the block image of the degree component.

Embodiment 6 FIG. FIG. 7 is a configuration diagram of the high-efficiency image decoder according to the present embodiment. The numbers in the figure are equivalent to the elements or signals of the numbers already described. Next, the operation of the device having the above configuration will be described. The decoder according to this embodiment is a decoder corresponding to the encoder according to the fifth embodiment. First, the input coded bit stream 115 is input to the demultiplexing unit 20, and is demultiplexed into a variable length coded word 109 of a luminance component, a variable length coded word 110 of a chrominance component, and a motion vector 136. Subsequently, each of the variable length decoding units 21a and 21a
b, the inverse quantization unit 7a, 7b, the chrominance component inverse transformation unit 8, and the luminance component inverse transformation unit 27, the decoded prediction error image 120 of the luminance component, and the decoded prediction error image 121 of the chrominance component Are input to the luminance / chrominance component combination blocker 30, and the prediction error image 132 is output according to the sample density ratio 219. 132 and the predicted image 135 are added in the adder 9d, and a decoded image 133 is output. When outputting the prediction image 135, the motion vector 136 is used as described above.

In the above embodiment, the combination of the luminance and chrominance components is performed on the decoder side at the specified sample density ratio included in the input coded bit stream to be decoded. However, without using the sample density ratio 219 sent from the encoder, the luminance / chrominance component combination blocking unit 3 according to an arbitrary sample density ratio at the decoder side.
At 0, the luminance component block and the chrominance component block may be combined. This corresponds to, for example, a case where the luminance / chrominance component combination blocker 30 does not have a blocking function according to the sample density ratio used in the encoder.

Embodiment 7 FIG. The motion compensation prediction unit will now be described in which the motion compensation prediction unit is further devised and the density ratio between luminance and color difference is considered. The local decoding unit in the fifth embodiment (FIG. 6) has a relatively simple configuration. FIG. 8 is a diagram showing the internal configuration of the motion compensation prediction unit shown in FIG. In the figure, 31 is a predicted image generation unit, 32 is a chrominance component prediction error calculation unit, 33 is a luminance component prediction error calculation unit, 34 is a weight coefficient W1 multiplication unit, 35
Is a weight coefficient W2 multiplier, 36 is an adder for the prediction error value,
Reference numeral 7 denotes a motion vector / predicted image determination unit, and reference numeral 38 denotes a luminance / color difference component separation unit. 140 is a predicted image obtained as a result of the motion search, 141 is a motion vector obtained simultaneously with 140, 142 is a predicted image of a luminance component, 143 is a predicted image of a chrominance component, 144 is a luminance component prediction error value, and 145 is a chrominance component The prediction error value, 146 is a weighted luminance component prediction error value, 147 is a weighted chrominance component prediction error value, 148 is an addition value of the prediction error values 146, 147, and 149 is separated from the input image 100. The luminance image 150 is a color difference image also separated. Other than the above, the numbers are the same as those already described.

Next, the operation of the apparatus having the above configuration will be described. The motion compensation prediction according to the present embodiment is characterized in that a prediction error value of a luminance component and a prediction error value of a chrominance component are multiplied by weighting factors W1 and W2, respectively, and the sum thereof is used as a prediction error value. . Thus, compared to the motion compensation based on the result of simply adding, by setting a weight of a luminance component that is visually more important than the color difference component, it is possible to obtain a motion compensation prediction that emphasizes the motion of the luminance component. . In the motion vector / predicted image determining unit 37, the motion vector when the minimum value is given among the added values 148 of the prediction errors calculated in all the search ranges is 136, and the predicted image is 135, and the motion compensated predicting unit 25a Output.

Embodiment 8 FIG. In the above embodiment, the high-efficiency encoder and the high-efficiency decoder have been described separately. These can be combined to construct a high-efficiency encoding / decoding system for images.

[0038]

As described above, the image encoder according to the present invention performs another blocking of the luminance component and the chrominance component in consideration of the sample density ratio of the luminance component and the chrominance component, and Since the components are separately encoded, the diffusion range of color noise can be reduced, and there is an effect of suppressing color distortion.

Further, the image decoder of the present invention separately decodes an input bit stream in which a luminance component and a chrominance component are separately encoded, and then combines the two components again by blocking. There is an effect that a decoded image with reduced color noise can be obtained.

[Brief description of the drawings]

FIG. 1 is a configuration block diagram of a high efficiency encoder according to Embodiment 1 of the present invention.

FIG. 2 shows a luminance pixel of an image format of 4: 2: 0.
It is a figure showing a color difference pixel.

FIG. 3 is a configuration block diagram of a high-efficiency decoder according to Embodiment 2 of the present invention.

FIG. 4 is a configuration block diagram of a high efficiency encoder according to Embodiment 3 of the present invention.

FIG. 5 is a block diagram illustrating a configuration of a high-efficiency decoder according to a fourth embodiment of the present invention.

FIG. 6 is a configuration block diagram of a high efficiency encoder according to Embodiment 5 of the present invention.

FIG. 7 is a configuration block diagram of a high-efficiency decoder according to Embodiment 6 of the present invention.

FIG. 8 is a configuration block diagram of a motion compensation prediction unit according to a seventh embodiment.

FIG. 9: Luminance pixels of 4: 2: 2 image format
It is a figure showing a color difference pixel.

FIG. 10 is a configuration block diagram of a conventional encoder.

FIG. 11 is a configuration block diagram of a conventional decoder.

[Explanation of symbols]

1. Luminance / chrominance component separation block unit, 2a, 2b subtractor, 3 luminance component conversion unit, 4 chrominance component conversion unit, 5a
Luminance signal quantization section, 5b chrominance signal quantization section, 6a
Luminance signal variable length coding section, 6b color difference signal variable length coding section, 7a luminance inverse quantization section, 7b color difference inverse quantization section, 8
Color difference component inverse transform unit, 9, 9a, 9b, 9c, 9d adder, 10 quantization control unit, 11 buffer, 12 multiplexing unit, 13 luminance component frame memory, 14 color difference frame memory, 15 luminance component motion compensation prediction Part 1
6 chrominance component motion compensation prediction unit, 20 demultiplexing unit, 2
1a luminance variable length decoding unit, 21b color difference variable length decoding unit, 22 color difference component motion compensation unit, 23 luminance component motion compensation unit, 24 frame memory, 25, 25a motion compensation prediction unit, 26 motion compensation prediction unit, 27 luminance Component inverse conversion unit, 30 luminance / chrominance component combination blocking unit, 31 predicted image generation unit, 32 color difference component prediction error calculation unit, 33
Luminance component prediction error calculation unit, 34 weight coefficient W1 multiplication unit,
35 weighting coefficient W2 multiplier, 36 prediction error value adder,
37 motion vector / predicted image determination unit, 50 A / D converter, 51 switch, 52 DCT conversion unit, 53 rate control unit, 54 quantization unit, 55 inverse quantization unit, 56
Inverse DCT section, 57 variable length coding section, 58 transmission buffer, 59 adder, 60 frame memory, 61 motion compensation prediction section, 62 reception buffer, 63 variable length decoding section, 64 format conversion section, 65 D / A conversion Unit, 100 digitized image data, 101
YC separated luminance component block image, 102 Y
C-separated block image of chrominance component, 103 prediction error image of block image of luminance component, 104 prediction error image of block image of chrominance component, 105 converter of luminance component, 106 conversion coefficient of chrominance component, 107 Quantization coefficient of luminance component, 108 Quantization coefficient of chrominance component, 1
09 Variable length coded word of luminance component, 110 Variable length coded word of chrominance component, 111 Variable length coded word of both luminance and chrominance components, 112 Variable length coded word sent from buffer, 113 Buffer output, 114 Adaptive quantization value, 1
15 coded bit stream, 116 motion vector of chrominance component, 117 motion vector of luminance component, 118
Dequantized transform coefficient of luminance component, 119 dequantized transform coefficient of chrominance component, 120 inverse-transformed prediction error image of 121 luminance component, 121 inverse-transformed prediction error image of chrominance component, 122 luminance component Locally decoded image of 1
23 Locally decoded image of chrominance component, 124 Predicted image of chrominance component read, 125 Predicted image of luminance component read, 126 Predicted image of chrominance component, 127 Predicted image of luminance component, 128 Combined block image, 129 motion prediction error image, 130 luminance component prediction error blocked image, 131 color difference component prediction error blocked image, 132 decoded prediction error image, 133 local decoded image, 134 frame memory Reference image read from 24, 135 motion prediction image, 136 motion vector, 140 quantization step size, 141 motion vector, 142 luminance component prediction image, 143 color difference component prediction image, 144 luminance component prediction error value, 145 Chrominance component prediction error value, 146 weighted luminance component prediction error value, 147 weighted Color difference component prediction error values, 148 146 and the sum of 147, the luminance separated from 149 input image 100, 15
0 separated color difference image, 200 input image, 201 digitized input image, 202 motion prediction error image, 20
3 Intra image or motion prediction error image, 204 D
CT transform coefficient, 205 adaptive quantized value, 206 quantized coefficient, 207 signal indicating information generation amount, 208 inverse-quantized DCT transform coefficient, 209 inverse DCT-processed intra image or prediction error image, 210 local decoded image ,
211 motion-compensated predicted image, 212 reference image read from the frame memory, 213 motion-compensated predicted image, 214 motion vector, 215 variable-length coded word,
216 coded bitstream, 217 format converted image, 218 D / A converted image output, 219 sample density ratio.

Claims (9)

[Claims] 1. An image encoder for encoding a target image signal and outputting an encoded bit stream, wherein the image signal is separated into a luminance signal and a chrominance signal, and is divided into blocks at a predetermined sample density ratio. A luminance / chrominance component separating unit, a luminance component converting unit for orthogonally transforming the separated luminance signal into a frequency domain, a luminance signal quantizing unit for quantizing the orthogonally transformed luminance signal, A luminance signal encoding unit that encodes the obtained luminance signal, a color difference component conversion unit that orthogonally transforms the separated color difference signal into a frequency domain, and a color difference signal quantization unit that quantizes the orthogonally transformed color difference signal. A chrominance signal encoding unit that encodes the quantized chrominance signal; and a multiplexing unit that multiplexes the luminance signal coded output and the chrominance signal coded output. High-efficiency coder for image, characterized in that the stream. 2. A luminance signal local decoding means for inversely quantizing the quantized luminance signal, further performing an inverse transform, and feedback-connecting the luminance signal after separation via a luminance predictor, and inversely transforming the quantized color difference signal. 2. A high-efficiency image encoder according to claim 1, further comprising a chrominance signal local decoding means for quantizing, further inversely transforming, and returning to the separated chrominance signal via a chrominance predictor. 3. A luminance inverse quantization / inverse transforming unit for inversely quantizing the quantized luminance signal and further performing inverse transform, and a color difference inverse quantizing / inverting unit for inversely quantizing the quantized color difference signal and further performing inverse transform.
An inverse transformation unit, a luminance / chrominance component combination blocking unit that combines the luminance signal after the inverse transformation and the chrominance signal after the inverse transformation into a block signal, and a luminance / chrominance signal after the integration via a predictor 2. A high-efficiency image encoder according to claim 1, wherein a local decoding means for performing subtractive feedback connection is added to a target image signal. 4. The predictor in the local decoding means includes a motion compensation prediction unit, wherein the motion compensation prediction unit separates the motion prediction of a luminance component and the motion prediction of a chrominance component from each other. 4. The high-efficiency image encoder according to claim 3, wherein an error with the color difference signal is calculated, the calculated prediction error is weighted, added, and subtracted and fed back to the target image signal. 5. An image decoder for obtaining an image signal from a coded bit stream to be decoded, comprising: a multiplexing / demultiplexing unit for separating the input coded bit stream into coded data of a luminance signal and a chrominance signal; A luminance decoding unit that decodes the separated luminance encoded data; a luminance inverse quantization unit that inversely quantizes the decoded luminance decoded signal; and an inverse orthogonalization of the inversely quantized luminance decoded signal. A luminance inverse transform unit for converting, a color difference decoding unit for decoding the separated color difference encoded data, a color difference inverse quantization unit for inverse quantizing the decoded color difference decoded signal, and the inverse quantization A chrominance inverse transform unit for performing an inverse orthogonal transform on the decoded chrominance decoded signal; and a luminance / chrominance component combining unit for combining the luminance signal output subjected to the luminance inverse orthogonal transform and the color difference signal output subjected to the chrominance inverse orthogonal transform. ·Color difference High-efficiency decoder for image, characterized in that the coupling output a reproduction image signal. 6. A luminance predicting means for predicting the luminance of the luminance signal subjected to the luminance inverse orthogonal transformation via a luminance adder and feedback-connecting the luminance signal after the prediction to the luminance adder; 6. A high-efficiency image according to claim 5, further comprising a color difference predicting means for predicting the color difference signal via said color difference adder and for connecting said color difference signal after said prediction back to said color difference adder. Decoder. 7. An output of a luminance / chrominance combining unit that combines a luminance signal subjected to luminance inverse orthogonal transformation and a color difference signal subjected to color difference inverse orthogonal transformation is predicted via an image adder, and the image signal after the prediction is calculated. 6. A high-efficiency image decoder according to claim 5, wherein prediction means for feedback connection to said image adder is added. 8. The image processing apparatus according to claim 5, wherein the luminance / chrominance combining unit combines the luminance and the chrominance according to a specified sample density ratio included in the coded bit stream to be decoded. High efficiency decoder. 9. A luminance / chrominance separation unit for separating a target image signal into a luminance signal and a chrominance signal to form a block, and a luminance for quantizing the luminance signal after the orthogonal transformation of the separated luminance signal into a frequency domain. A signal quantization unit, a luminance signal encoding unit that encodes the quantized luminance signal, and a chrominance signal quantization unit that quantizes the chrominance signal after the orthogonal transform of the separated chrominance signal into a frequency domain. A color difference signal encoding unit that encodes the quantized color difference signal, and a high-efficiency image encoder that includes a multiplexing unit that multiplexes the luminance signal encoded output and the color difference signal encoded output. A multiplexing / demultiplexing unit that separates an encoded bit stream from the image high-efficiency encoder into encoded data of a luminance signal and a chrominance signal, and a luminance decoding unit that decodes the separated encoded luminance data. , The decoded luminance A luminance inverse quantization unit for inversely quantizing the signal, a color difference decoding unit for decoding the separated color difference encoded data, and a color difference inverse quantization unit for inverse quantizing the decoded color difference decoded signal. A luminance / chrominance combining unit that combines a luminance signal output after inverse orthogonal transformation of the inversely quantized luminance decoded signal and a color difference signal output after inverse orthogonal transformation of the inversely quantized color difference decoded signal. High-efficiency encoding / decoding system for images, comprising a high-efficiency decoder for images.