JP2003250161A - Encoder and decoder - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an encoder which encodes image data according to predictive coding including motion detection, and increases motion detection precision while suppressing an increase in the amount of data processed at the predictive coding. <P>SOLUTION: The encoder includes: a 4×4 block divider 101 for dividing image data corresponding to a target frame to be processed into 4×4 pixel blocks; a motion compensation unit 111 for performing motion compensation for the image data in units of 4×4 pixel blocks thereby to generate predicted data; and an 8×8 pixel block configuration unit 103 for transforming difference data between image data corresponding to the 4×4 pixel block and predicted data corresponding to the 4×4 pixel block into prediction error data of an 8×8 pixel block. This encoder carries out a DCT process, a quantization process, and a variable length coding process for prediction error data, in units of 8×8 pixel block. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding device and a decoding device, and more particularly to a device for coding a video signal using inter-frame prediction and a coded video signal for decoding using inter-frame prediction. The present invention relates to a device for converting the information.

[0002]

2. Description of the Related Art Conventionally, as a method for efficiently compressing and coding a digital format video signal, ISO / IEC (Internation
al Organization for Standardization / Internationa
l Electrotechnical Commission) 13818-2 (MPEG
(Moving Picture Experts Group) -2 video) or M such as ISO / IEC 11172-2 (MPEG-1 video)
The PEG coding method is generally used.

The MPEG encoding system is composed of an intra-frame encoding process for encoding image data using intra-frame correlation of pixel values and an inter-frame encoding process for encoding image data using inter-frame correlation of pixel values. The encoding process and the encoding process are adaptively switched. In this MPEG encoding method, the encoding process of the image data corresponding to a moving image is performed with the image data corresponding to a plurality of consecutive frames as one unit. Here, a plurality of consecutive frames that are units of encoding processing are group of pictures (GO).
P). Specifically, in this MPEG encoding method, an intraframe encoding process is performed on at least one frame of the plurality of frames forming this GOP, and an interframe encoding process is performed on the remaining frames. Processing is performed.

The interframe coding process includes two processes, a forward interframe predictive coding process and a bidirectional interframe predictive coding process. A frame to which the forward interframe predictive coding process is applied is called a P frame, and a frame to which the bidirectional interframe predictive coding process is applied is B.
Called a frame. The P frame is a frame (reference frame) located before the P frame on the display time axis.
The predictive coding process is performed with reference to. Predictive coding processing is performed on the B frame by referring to two frames (reference frames) located close to and before and after the B frame on the display time axis. Normally, in the P frame encoding process, an I frame close to the P frame is used as a reference frame, and in the B frame encoding process,
An I frame and a P frame adjacent to the B frame, or two P frames are used as reference frames.

FIG. 16 is a diagram for specifically explaining a plurality of frames (FIG. (A)) constituting one GOP and coded data (FIG. (B)) corresponding to the GOP. FIG.
Shows a plurality of consecutive frames F (k-5) to F (k + 12) and coded data D (k-5) to D (k + 12) corresponding to each frame in comparison. There is. Note that k is an arbitrary integer.
The frames F (k-5) to F (k + 12) are arranged in display time order, while the data D (k- corresponding to the frames F (k-5) to F (k + 12) are arranged. 5) to D (k + 12) are arranged in coding time order (in other words, decoding time order).

Specifically, F (k) and F (k + 12) are intra-frame coded frames (I frames), and F (k-3),
F (k + 3), F (k + 6), F (k + 9) are forward inter-frame predictive coding frames (P frames), F (k-5), F (k-4), F ( k
-2), F (k-1), F (k + 1), F (k + 2), F (k + 4), F (k + 5),
F (k + 7), F (k + 8), F (k + 10), and F (k + 11) are bidirectional interframe predictive coding frames (B frames). D (k) and D (k + 12) are I-frame encoded data, and D (k-3), D (k + 3), D (k + 6), and D (k + 9). ) Is the encoded data of P frame, D (k-5), D (k-4), D (k-2), D (k-
1), D (k + 1), D (k + 2), D (k + 4), D (k + 5), D (k + 7),
D (k + 8), D (k + 10), D (k + 11) are coded data of the B frame.

Here, one G is composed of 12 frames from B frame F (k-2) to P frame F (k + 9).
OP is configured. For example, P frame F (k + 3) is subjected to interframe predictive coding processing using I frame F (k) as a reference frame. Further, the P frame F (k + 6) is subjected to interframe predictive coding processing using the P frame F (k + 3) as a reference frame.
Further, for B frames F (k + 1) and F (k + 2), interframe predictive coding processing is performed using the I frame F (k) and P frame F (k + 3) as reference frames. Is given.

For the coded data corresponding to each frame obtained by the above coding process, in order to reduce the capacity of the memory used in the decoding process, the array of the coded data of each frame is arranged. To convert the array according to the display order of the images of each frame (array in display time order) to the array according to the order of performing decoding processing of each frame (array in decoding time order) (array conversion processing) ) Is given.

Specifically, in the encoded data Ed obtained by subjecting the encoded data corresponding to the GOP to the array conversion processing, as shown in FIG.
The coded data D (k) of (k) is located at the head of the coded data Dgop corresponding to the GOP, and then the B frame F (k-
2) coded data D (k-2), B frame F (k-1) coded data D (k-1), P frame F (k + 3) coded data D
(k + 3) follows in order. Then, the encoded data Dgop corresponding to the GOP is transmitted or recorded in the order after the array conversion processing.

FIG. 17 is a block diagram for explaining a conventional MPEG encoding device. This encoding device 700
Is for performing interframe coding processing of the MPEG coding system on an input video signal (hereinafter, also referred to as image data) Id. That is, the encoding device 700 has 16 pixels in the horizontal direction for the image data of each frame with respect to the input image data Id,
A 16 × 16 block divider 701 that performs a dividing process so as to correspond to a 16 × 16 pixel block (macroblock) having 16 pixels in the vertical direction and outputs image data 721 corresponding to each macroblock; 16x above
A motion compensation unit 711 that performs a motion detection process for each macroblock on the output data 721 of the 16-block divider 701 and outputs a motion vector 724 and prediction reference data (data of a prediction reference block) 726 for the target block. Have

Here, the motion detection is a process for detecting an area in the predicted reference frame in which the image data is stored in the reference frame memory 710, which most closely matches the pixel level pattern corresponding to the target block. A matching method is used. The region in the prediction reference frame thus detected by motion detection is the prediction reference block, and the information indicating the position of the prediction reference block in the prediction reference frame is the motion vector 7
24.

The encoding device 700 uses the data 721 of the target block to the data 726 of the prediction reference block.
Subtractor 70 that performs subtraction processing for subtracting and outputs difference data of both blocks as prediction error data (data of prediction error block) 727 of 16 × 16 pixel block
Have two. In the subtraction process, the pixel level of each pixel of the target block and the pixel level of the pixel corresponding to each pixel of the target block of the prediction reference block are subtracted for each pixel.

The encoding device 700 includes the subtractor 70.
For the image data 727 corresponding to the 16 × 16 pixel block (macroblock) which is the output of 2, the image data of each macroblock has 8 pixels in the horizontal direction,
An 8 × 8 block configuration unit 703 is provided, which performs a dividing process corresponding to an 8 × 8 pixel block having eight pixels in the vertical direction and outputs prediction error data 722 corresponding to each 8 × 8 pixel block. is doing.

That is, in the 8 × 8 block construction unit 703, the prediction error data 727 of the 16 × 16 pixel block (16 × 16 prediction error block) BLp is converted into four 8 × 8 pixel blocks ( (8 × 8 prediction error block) divided into prediction error data 722 of BLpa, BLpb, BLpc, BLpd.

The above-mentioned encoding device 700 applies a two-dimensional discrete cosine transform (D) to the prediction error data 722 of the 8 × 8 pixel block output from the 8 × 8 block composer 703.
CT) processing and D corresponding to each 8 × 8 pixel block
A DCT unit 704 which outputs a CT coefficient 714,
Each DCT coefficient of each 8 × 8 pixel block, which is the output 714 of the converter 704, is quantized by a predetermined quantization step to obtain each 8
And a quantizer 705 that outputs a quantized coefficient 723 corresponding to a × 8 pixel block.

The encoding device 700 scans the quantized coefficients 723 of each 8 × 8 pixel block having a two-dimensional array into a data string having a one-dimensional array by scanning along a predetermined path, It has a variable length encoder 706 which encodes the one-dimensional data sequence by variable length encoding and outputs an encoded data sequence Bstr. The variable length encoder 706 also outputs the information of the motion vector 724 obtained by the motion compensation unit 711, and the output of the encoded data string Bstr is the motion vector 7
24 pieces of information are also added.

The encoding device 700 includes the quantizer 7
The inverse quantizer 707 that performs the inverse quantization process on the quantized coefficient 723 of each 8 × 8 pixel block, which is the output of 05, and restores the DCT coefficient of each 8 × 8 pixel block, and the inverse quantizer 707. Inverse D which obtains decoded prediction error data of each 8 × 8 pixel block (decoded prediction error block) by performing inverse DCT processing on the DCT coefficient which is the output 715 of the digitizer 707.
Receiving the output 728 of the CT unit 708 and the inverse DCT unit 708, combining the decoded prediction error data 728 of four 8 × 8 pixel blocks (8 × 8 decoded prediction error block),
16 × 1 for generating decoded data 729 corresponding to a 16 × 16 pixel block (16 × 16 decoded prediction error block)
6 block construction unit 712.

The image encoding apparatus 700 is 16 × 16.
The decoded prediction error data 729 of the pixel block (16 × 16 decoded prediction error block) and the 16 × 16 pixel block (1
An adder 709 that adds the prediction reference data 726 of the 6 × 16 prediction reference block) and outputs the decoded data 730 of the 16 × 16 pixel block (16 × 16 decoding block)
And a reference frame memory 71 for storing the output 730 of the adder 709 as image data of a reference frame used in predictive coding for the next frame of the target frame.
It has 0 and. The addition processing in the adder 709 is performed by the pixel level of each pixel of the 16 × 16 decoded prediction error block and the 16 × 16 prediction reference block.
The addition with the pixel level of the pixel corresponding to each pixel of the × 16 decoded prediction error block is performed for each pixel.

Next, the operation will be described. A frame to be processed (target frame) by the encoding device 700.
When the image data Id is input, the 16 × 16 block divider 701 divides the image data Id so as to correspond to a 16 × 16 pixel block (macro block),
Image data 721 corresponding to each 16 × 16 pixel block
Is output.

Image data 7 of the 16 × 16 pixel block
When 21 is input to the motion compensating unit 711, the motion compensating unit 711 performs motion detection on the image data 721 of the 16 × 16 pixel block (target block) that is the processing target and corresponds to the target block. Motion vector 72
4 and prediction reference data (image data of 16 × 16 prediction reference block) 726 are output. This motion detection is to detect the region that most matches the pixel level pattern of the 16 × 16 pixel block in the predicted reference frame stored in the reference frame memory 710, and is performed by, for example, the block matching method. The area detected by this motion detection corresponds to the target block 16
The information indicating the position of the 16 × 16 prediction reference block in the reference frame is the motion vector 724 corresponding to the target block.

The subtractor 702 performs a subtraction process for subtracting the image data 726 of the 16 × 16 prediction reference block from the image data 721 of the target block, and the prediction error data of the 16 × 16 pixel block (of the 16 × 16 prediction error block). Image data) 727 is output.

Then, the 8 × 8 block construction unit 703 is
Prediction error data 727 of the 16 × 16 pixel block
Are the four 8 × that form the 16 × 16 pixel block.
It is divided so as to correspond to an 8-pixel block, and prediction error data of the 8 × 8 pixel block (image data of the 8 × 8 prediction error block) 722 is output. That is, the data 727 of the 16 × 16 prediction error block BLp is, as shown in FIG. 18, four 8 × 8 prediction error blocks BLpa, BLpb,
It is divided into data 722 of BLpc and BLpd.

When the data 722 of the 8 × 8 prediction error block is input to the DCT unit 704, the DCT unit 704 outputs 8
A two-dimensional discrete cosine transform (DCT) process is performed on the data 722 of the × 8 prediction error block to generate the DCT coefficient 714 for the 8 × 8 prediction error block.
The quantizer 705 determines the DC of this 8 × 8 prediction error block.
The T coefficient 714 is quantized in a predetermined quantization step,
The quantized coefficient 723 for the 8 × 8 prediction error block is generated.

The variable length encoder 706 scans a plurality of quantized coefficients 723 having a two-dimensional array for the 8 × 8 prediction error block along a predetermined path and converts them into a data string having a one-dimensional array. The array conversion process is performed, the data string is encoded by the variable length code, and the encoded data string Bstr is output. The variable length encoder 706 also adds the information of the motion vector 724 obtained by the motion compensating unit 711 to the encoded data string Bstr and outputs it.

The dequantizer 707 dequantizes the quantized coefficient 723 of the 8 × 8 prediction error block into 8 by dequantizing.
The DCT coefficient 715 of the × 8 prediction error block is restored, and the inverse DCT unit 708 performs a two-dimensional inverse DCT transform process on the DCT coefficient 715 of the 8 × 8 prediction error block to obtain 8
The decoded image data 728 of the × 8 prediction error block is output.

When the decoded image data 728 of the 8 × 8 prediction error block is input to the 16 × 16 block construction unit 712, the 16 × 16 block construction unit 712 decodes the decoded images of the four 8 × 8 prediction error blocks. The process of forming the decoded image data of one 16 × 16 prediction error block by combining the data 728 is performed, and the decoded image data 729 of the 16 × 16 prediction error block is output.

The adder 709 performs an addition process of adding the decoded image data 729 of the 16 × 16 prediction error block and the image data 726 of the 16 × 16 prediction reference block to decode the 16 × 16 pixel block. Image data 730 is generated.

The decoded image data 730 corresponding to the 16 × 16 pixel block (macroblock) is used as the image data of the reference frame used in the predictive coding process for the frame next to the target frame. It is stored in the memory 710.

Next, a conventional MPEG-4 decoding process for decoding the encoded data string output from the above encoding device will be described. FIG. 19 is a block diagram showing a conventional decoding device 800. This decoding device 800 is shown in FIG.
The encoded data string Bstr output from the encoding device 700 shown in FIG.

That is, the decoding apparatus 800 converts the input coded data string Bstr into the quantized coefficient code string 8
22 and the motion vector 830, and outputs the code classifier 801 and the code string 822 of the quantized coefficient, which is subjected to the variable length decoding process to obtain the quantized coefficient 823 of the 8 × 8 prediction error block. Variable length decoder 802 for sequentially restoring
And referring to the image data of the decoded frame stored in the reference frame memory 807, the image data of the area indicated by the motion vector 830 in the frame is 16
The motion compensation unit 806 outputs the data 831 of the × 16 prediction reference block.

The decoding device 800 performs inverse quantization processing on the quantized coefficient 823 of the 8 × 8 prediction error block using the quantization step used at the time of encoding to obtain 8
The inverse quantizer 803 that restores the DCT coefficient 814 of the 8 × 8 prediction error block and the DCT coefficient 814 of the 8 × 8 prediction error block is subjected to the inverse two-dimensional discrete cosine transform processing to obtain the 8 × 8 prediction error block. The inverse DCT unit 809 that generates the decoded image data 824 is included.

The decoding device 800 combines the decoded image data 824 of four 8 × 8 prediction error blocks to obtain the decoded image data 8 of one 16 × 16 prediction error block.
25 is performed to obtain the decoded image data 825 of the 16 × 16 prediction error block, the decoded image data of the 16 × 16 prediction error block, and the motion compensation unit 806. Given 16
It has an adder 805 which performs a process of adding the image data 831 of the × 16 prediction reference block and outputs the decoded image data 840 corresponding to the 16 × 16 pixel block.

The decoding device 800 restores the image data corresponding to the target frame from the image data 840 corresponding to the 16 × 16 pixel block, and outputs the restored data as the reproduced image data RId.
And decoded image data 840 of a 16 × 16 pixel block
Is stored as image data of a reference frame used in the predictive decoding process for the frame next to the processing target frame.

Next, the operation will be described. When the encoded data string Bstr output from the encoding device 700 is input to the decoding device 800, the code classifier 801
Is the coded data string Bstr and the quantized coefficient code string 82.
2 and the motion vector 830 are processed, and the code classifier 801 outputs a quantized coefficient code string 822.
And the motion vector 830 are output.

The variable length decoder 802 converts the quantized coefficient code string 822 into the quantized coefficient 823 of the 8 × 8 prediction error block by the variable length decoding process. Inverse quantizer 8
03 is the quantized coefficient 82 of this 8 × 8 prediction error block.
3 is inversely quantized by the quantization step used in the encoding process in the encoding device 700, and 8
The DCT coefficient 814 of the × 8 prediction error block is restored.
The inverse DCT unit 809 determines the D of this 8 × 8 prediction error block.
The CT coefficient 814 is transformed into the decoded image data 824 of the 8 × 8 prediction error block by the inverse two-dimensional discrete cosine transform process.

The 16 × 16 block construction unit 804 performs a process of combining decoded image data 828 of four 8 × 8 prediction error blocks to generate decoded image data of one 16 × 16 prediction error block, The decoded image data 825 of the × 16 prediction error block is output.

The motion compensating unit 806 also generates image data of the area indicated by the motion vector 830 in the frame by motion compensation referring to the image data of the decoded frame stored in the reference frame memory 807. The image data is output as data 831 of the 16 × 16 prediction reference block.

Then, the adder 805 adds the decoded image data 806 of the 16 × 16 prediction error block to the motion compensation unit 8
The process of adding the image data of the 16 × 16 prediction reference block 831 obtained in 06 is performed to generate and output the decoded image data 840 corresponding to the 16 × 16 pixel block.

The image configurator 808 generates image data corresponding to a frame from the decoded image data 840 corresponding to the 16 × 16 pixel block, and outputs the data as reproduced image data RId.

In addition, 1 output from the adder 805 is output.
Decoded image data 840 corresponding to a 6 × 16 pixel block
Is stored in the reference frame memory 807 as data of a reference frame to be referred to in the predictive decoding process for the frame next to the target frame.

[0041]

[Patent Document 1] US Pat. No. 4,821,119

[0042]

However, the conventional MP
The EG encoding method has a problem that when the image size to be handled is small, it is impossible to detect a fine motion, which causes a decrease in prediction efficiency.

Briefly described, in the conventional MPEG encoding system, as described above, motion compensation is performed in units of blocks (macroblocks) each consisting of 16 × 16 pixels.
DCT for each block (sub-block) consisting of × 8 pixels
Processing and quantization processing are performed. The image size handled by MPEG is NTSC (National Television Sta
ndards Committee) TV format is horizontal 72
It is generally 0 pixels and 480 vertical lines (720 × 480 pixels).

Therefore, in the case of the NTSC TV format, the image size of one frame is 720 × 480.
Since pixels are generally used, it is not a big problem to make a unit for performing motion compensation a macroblock consisting of 16 × 16 pixels. However, when the image size of one frame is small, for example, the image size of one frame is 276 x 1
In the case of about 44 pixels, the relative size of the macroblock with respect to one frame is about four times the relative size of the macroblock with respect to the frame having an image size of 720 × 480 pixels.

Therefore, when the image size of one frame is small, only rough motion can be detected by performing motion compensation in units of 16 × 16 pixels (macroblocks), which causes deterioration of prediction efficiency. There is a problem that it ends up. Further, in compression encoding of image data, the larger the unit (DCT block) in which DCT processing is performed, the more significant coefficients are concentrated in low frequency components, so that the compression efficiency of image data can be improved. There is also a problem that increasing the DCT block causes an increase in the amount of DCT processing.

In addition, P102-103 (Chapter 5 Concept of transform coding 5-2 Orthogonal transform) of "Introduction to Advanced Technology" series published by Ohmsha, Inc. (Image Information Compression Television Society, edited by Hiroshi Harashima, H3.8.25) Features and comparison (Fig. 2.
6)), the DCT block size and coding efficiency are described.

The present invention has been made in view of the above problems, and it is possible to suppress the increase in the data processing amount in the predictive coding process or the predictive decoding process while improving the motion detection accuracy. It is an object to provide a method and a decoding method, and an encoding device and a decoding device.

[0048]

An encoding apparatus according to the present invention (Claim 1) is an encoding apparatus for predictively encoding image data frame by frame using pixel value correlation between frames. A prediction process of predicting image data corresponding to a target frame, a data predicting unit that generates predicted image data by performing a prediction block including a plurality of pixels that partition the frame, and image data of the processing target frame, It is characterized in that it comprises a data coding unit for coding for each coding block composed of a plurality of prediction blocks, using prediction image data corresponding to the coding block.

According to the present invention (claim 2), in the encoding device according to claim 1, the data predicting section generates a predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes. Of the prediction image data corresponding to a plurality of types of prediction blocks of different shapes, select the prediction image data of the prediction block that has the highest coding efficiency for the coding block, to the coding block It is characterized by generating corresponding predicted image data.

According to the present invention (claim 3), in the encoding device according to claim 2, the data predicting unit predicts which shape of predictive image data corresponding to a plurality of types of predictive blocks having different shapes. It is characterized in that block identification information indicating whether the predicted image data of a block has been selected is output.

According to the present invention (claim 4), in the encoding device according to claim 1, the data predicting unit generates a predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes. The prediction image data corresponding to a plurality of types of prediction blocks having different shapes is arranged such that the arrangement pattern of the various prediction blocks in the encoding block is the arrangement pattern that maximizes the encoding efficiency for the encoding block. It is characterized in that predictive image data corresponding to the coded block is generated by such combination.

According to a fifth aspect of the present invention, in the encoding device according to the fourth aspect, the data prediction section outputs block arrangement information indicating an arrangement pattern of various prediction blocks in the encoded block. It is characterized by that.

An encoding apparatus according to the present invention (claim 6) is
In a coding device for predictively coding image data for each frame using inter-frame pixel value correlation, a prediction process for predicting image data corresponding to a frame to be processed is performed by a first pixel consisting of a plurality of pixels for partitioning the frame. A first data prediction unit that performs first prediction image data by performing each prediction block, and image data of the processing target frame for each first coding block including a plurality of first prediction blocks,
A first predictive encoding unit having a first data encoding unit that encodes using first predictive image data corresponding to the first encoded block, and predicts image data corresponding to a processing target frame A second data prediction unit that performs the prediction process to generate second predicted image data for each second prediction block composed of a plurality of pixels that partition the frame; A second prediction code having, for each second coding block that divides the two prediction blocks, a second data coding unit that codes using the second prediction image data corresponding to the second coding block. Depending on the conversion unit and the command signal from the outside,
The present invention is characterized by including a selecting unit that supplies one of the first predictive coding unit and the second predictive coding unit.

The encoding method according to the present invention (claim 7) is
A coding method for predictively coding image data for each frame by using pixel value correlation between frames, the predicting process for predicting image data corresponding to a frame to be processed is performed from a plurality of pixels that partition the frame. And a prediction step of generating predicted image data for each prediction block, and the image data of the processing target frame is used for each coding block including a plurality of prediction blocks using the prediction image data corresponding to the coding block. And an encoding step for encoding.

According to the present invention (claim 8), in the coding method according to claim 7, the predicting step includes a step of generating predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes, and Of the prediction image data corresponding to a plurality of types of prediction blocks having different shapes, a step of selecting the prediction image data of the prediction block having the highest coding efficiency for the coding block, and the prediction image data of the selected prediction block According to the above, a step of generating predicted image data corresponding to the coded block is included.

According to the present invention (claim 9), in the encoding method according to claim 8, the predicting step includes a predictive block of any shape among predictive image data corresponding to a plurality of kinds of predictive blocks having different shapes. It is characterized by including a step of generating block identification information indicating whether or not the predicted image data has been selected.

According to the present invention (claim 10), in the encoding method according to claim 7, the predicting step includes a step of generating predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes; The step of determining the arrangement pattern of the various prediction blocks in the coding block so that the coding efficiency with respect to the coding block is highest, and the prediction image data corresponding to a plurality of types of prediction blocks having different shapes are Generating the predicted image data corresponding to the coding block by combining the arrangement patterns of the various prediction blocks in the encoded block so as to be the determined arrangement pattern.

According to the present invention (claim 11), in the encoding method according to claim 10, the prediction step includes a step of generating block arrangement information indicating an arrangement pattern of various prediction blocks in the encoded block. It is characterized by that.

A decoding device according to the present invention (claim 12) is a decoding device for predictively decoding coded data obtained by predictively coding image data, for each frame using pixel value correlation between frames. A device, which is a device, performs a prediction process of predicting image data corresponding to a frame to be processed for each prediction block including a plurality of pixels that divide the frame, and a data prediction unit that generates predicted image data; A data decoding unit that decodes encoded data of a frame for each decoding block including a plurality of prediction blocks by using predicted image data corresponding to the decoding block is provided. is there.

According to the present invention (claim 13), in the decoding device according to claim 12, the data predicting unit generates a predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes. In predictive encoding of the image data, block identification information indicating which shape of the prediction image data of the prediction image data of the prediction image data corresponding to the plurality of types of prediction blocks having different shapes is selected. On the basis of the above, the predictive image data corresponding to the predictive block having a predetermined shape is selected, and the predictive image data corresponding to the decoding block is generated by the predictive image data of the selected predictive block. To do.

According to the present invention (claim 14), in the decoding device according to claim 12, the data prediction unit generates a predicted image data corresponding to each of a plurality of types of prediction blocks having different shapes. Corresponding to the decoding block by combining prediction image data corresponding to a plurality of types of prediction blocks having different shapes based on block arrangement information indicating an arrangement pattern of various prediction blocks in the decoding block. It is characterized by generating predicted image data.

A decoding apparatus according to the present invention (Claim 15) is a decoding apparatus for predictively decoding coded data obtained by predictively coding image data, for each frame using pixel value correlation between frames. A first device that performs prediction processing for predicting image data corresponding to a frame to be processed for each first prediction block including a plurality of pixels that partition the frame, and generates first prediction image data. Decoding the data prediction unit and the encoded data of the processing target frame for each first decoding block including a plurality of first prediction blocks using the first prediction image data corresponding to the first decoding block A first predictive decoding unit having a first data decoding unit for converting into a second predictive decoding unit including a plurality of pixels partitioning the frame, and a predictive process of predicting image data corresponding to a frame to be processed. A second data predicting unit that performs second prediction image data by performing each block, and coded data of the processing target frame for each second decoding block that divides the second prediction block A second predictive decoding unit having a second data decoding unit that decodes using the second predictive image data corresponding to the two decoding blocks, and the above-mentioned encoding according to a command signal from the outside. The present invention is characterized by including a selecting unit that supplies data to one of the first predictive decoding unit and the second predictive decoding unit.

A decoding method according to the present invention (claim 16) is a decoding method for predictively decoding coded data obtained by predictively coding image data for each frame by using pixel value correlation between frames. A prediction step of performing prediction processing of predicting image data corresponding to a frame to be processed for each prediction block composed of a plurality of pixels for partitioning the frame, and a frame to be processed, And a decoding step of decoding the coded data of (1) for each decoding block including a plurality of prediction blocks using the prediction image data corresponding to the decoding block.

According to the present invention (claim 17), in the decoding method according to claim 16, the predicting step includes a step of generating predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes; At the time of predictive coding of image data, a predetermined value is determined based on block identification information indicating which shape of the predictive image data is selected from the predictive image data corresponding to a plurality of types of predictive blocks having different shapes. And a step of generating predicted image data corresponding to the decoded block from the predicted image data of the selected predicted block. It is a thing.

According to the present invention (claim 18), in the decoding method according to claim 16, the predicting step generates predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes, and Prediction image data corresponding to a plurality of types of prediction blocks having different shapes are combined based on block arrangement information indicating an arrangement pattern of various prediction blocks in the decoding block to generate prediction image data corresponding to the decoding block. And a step of performing.

[0066]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below. (Embodiment 1) FIG. 1 is a block diagram for explaining an encoding apparatus according to Embodiment 1 of the present invention. In the coding apparatus 100a of the first embodiment, the data unit for motion compensation is smaller than the data unit for DCT processing.

That is, the encoding apparatus 100a is similar to that shown in FIG.
Unlike the 16 × 16 block divider 701 of the conventional encoding apparatus 700 shown in FIG. 7, the image data of each frame input as a video signal is 4 × 4 in which the number of pixels in the vertical and horizontal directions is 4, respectively. Unlike the first 4 × 4 block divider 101 that divides to correspond to a pixel block and the motion compensator 711 of the encoding device 700, 4 × 4.
Motion detection is performed for each pixel block, and a motion vector 124 for each 4 × 4 pixel block and prediction data (hereinafter also referred to as 4 × 4 prediction reference block data) 126 for each 4 × 4 pixel block are output. Motion compensation section 111.

The encoding apparatus 100a uses the data 121 of the 4 × 4 pixel block which is the output 121 of the 4 × 4 block divider 101 to the data 126 of the 4 × 4 prediction reference block.
To a subtractor 102 that outputs difference data (hereinafter, also referred to as 4 × 4 prediction error block data) 127 corresponding to the 4 × 4 pixel block, and an output 127 of the subtractor 102. And an 8 × 8 block configurator 103 for generating difference data (hereinafter, 8 × 8 prediction error block data) 122 corresponding to an 8 × 8 pixel block composed of four 4 × 4 pixel blocks. There is.

The encoding apparatus 100a performs the DCT process on the data 122 of the 8 × 8 prediction error block D.
The CT unit 104, the quantizer 105 that performs a quantization process on the output 114 of the DCT unit 104 with a predetermined quantization step, and the variable length coding process on the output 123 of the quantizer 105. Then, the code of the motion vector 124 from the motion compensating unit 111 is added to the coded data obtained by the variable length coding process, and the coded data string B is added.
The variable length encoder 106 that outputs s1.

The encoding apparatus 100a performs an inverse quantizer 107 on the output 123 of the quantizer 105 and an inverse DCT process on the output 115 of the inverse quantizer 107. And an inverse DCT unit 108 for generating decoded data 128 of 8 × 8 prediction error block.

The encoding apparatus 100a is the same as that shown in FIG.
Unlike the 16 × 16 block divider 701 of the encoding apparatus 700 shown in FIG. 7, the decoded data 128 of the 8 × 8 prediction error block is divided into four 4 × 4 prediction error blocks that partition the 8 × 8 prediction error block. And a second 4 × 4 block divider 112 that outputs the decoded data 129 of the 4 × 4 prediction error block.

Further, the encoding device 100a is 4 × 4.
The decoded data 129 of the prediction error block and the data 126 of the 4 × 4 prediction reference block are added to obtain decoded data corresponding to a 4 × 4 pixel block (hereinafter, also referred to as 4 × 4 decoded block data) 130. Adder 1 that outputs
09 and a reference frame memory 110 for storing the data 130 of the 4 × 4 decoded block as the data of the reference frame to be referred to at the time of predictive encoding for the frame next to the processing target frame.

The coding apparatus 10 according to the first embodiment
0a DCT device 104, quantizer 105, inverse quantizer 1
07, the inverse DCT unit 108, and the variable-length encoder 106 are respectively D in the conventional encoding device 700 shown in FIG.
The CT unit 704, the quantizer 705, the inverse quantizer 707, the inverse DCT unit 708, and the variable length encoder 706 are exactly the same.

Next, the operation will be described. When a video signal is input as image data Id to the encoding device 100a according to the first embodiment, the first 4 × 4 block divider 101 outputs the image data Id corresponding to the frame to be processed. The data 121 of the 4 × 4 pixel block is output after being divided so as to correspond to the 4 × 4 pixel block. The motion compensation unit 111 uses the data 121 of the 4 × 4 pixel block.
The motion vector 124 corresponding to the 4 × 4 pixel block to be processed by referring to the image data of the reference frame stored in the reference frame memory 110, and the 4 × 4 pixel Prediction data (4 × 4 prediction reference block data) 126 corresponding to the block
Is output. In the first embodiment, the motion compensation unit 111
The motion detection in 1 is different from the motion detection in the motion compensation unit 711 of the conventional encoding device 700 in that it is performed in units of 4 × 4 pixel blocks.

The subtractor 102 uses the data 121 of the 4 × 4 pixel block to the data 126 of the 4 × 4 prediction reference block.
Is performed, and difference data (4 × 4 prediction error block data) 127 corresponding to the 4 × 4 pixel block obtained by the subtraction process is output. The subtraction process is a process of subtracting the pixel levels of the corresponding pixels of the 4 × 4 pixel block and the 4 × 4 prediction reference block.

Data 12 of the 4 × 4 prediction error block
When 7 is input to the 8 × 8 block construction unit 103, data 122 of 8 × 8 prediction error block is generated from four 4 × 4 prediction error block data 127.

FIG. 2 is a diagram showing the relationship between a frame and each block. Here, the size of one frame F,
That is, the number of pixels in the horizontal and vertical directions of the frame F is 3
It is 2. For example, as shown in FIG. 2 (a), the first 4 ×
The data of the four 4 × 4 pixel blocks BL4a to BL4d obtained by the four-block divider 101 is the frame F.
Corresponds to the data of one 8 × 8 pixel block BL8. Also, each 4 × 4 pixel block BL4a to BL4d
Is a 4 × 4 prediction error block BLp4a to BLp4d (see FIG. 2).
It corresponds to (b)), and one 8 × 8 prediction error block BLp8 is formed from these four 4 × 4 prediction error blocks BLp4a to BLp4d.

Data 12 of the above 8 × 8 prediction error block
2 is input to the DCT device 104, the DCT device 104
Is 2 for the data 122 of the 8 × 8 prediction error block.
D, which has been subjected to a three-dimensional discrete cosine transform (DCT) process and which is composed of a plurality of DCT coefficients corresponding to 8 × 8 prediction error blocks
The CT coefficient data 114 is output. The quantizer 105
Based on the DCT coefficient data 114, each DCT coefficient of the 8 × 8 prediction error block is quantized in a predetermined quantization step, and the quantized coefficient data 123 including a plurality of quantized coefficients corresponding to the 8 × 8 prediction error block is quantized. Output. This 8
The plurality of quantized coefficients corresponding to the × 8 prediction error block is 2
It has an array of dimensions.

Based on the quantized coefficient data 123, the variable-length encoder 106 quantizes a plurality of quantized coefficients 123 having a two-dimensional array corresponding to the 8 × 8 prediction error block into a one-dimensional array. A variable length coding process is performed in which scanning is performed so as to be converted into a coefficient, and the quantized coefficient having the one-dimensional array obtained by the scanning is coded using a variable length code. At this time, the variable length encoder 106
Motion vector 124 obtained by motion compensation unit 111
Is also encoded. Then, the variable length encoder 106
The motion vector 124 together with the encoded data of the quantized coefficient
The encoded data string Bs1 including the code of is output.

When the quantized coefficient data 123 of the 8 × 8 prediction error block is input to the inverse quantizer 107, the inverse quantizer 107 uses the quantization step used in the quantizer 105 to Inverse quantization processing is performed on the quantized coefficient data 123, and a plurality of D of 8 × 8 prediction error blocks
The DCT coefficient data 115 composed of CT coefficients is restored.
When the DCT coefficient data 115 of the 8 × 8 prediction error block is input to the inverse DCT device 108, the inverse DCT device 108
The DCT coefficient data 115 of the 8 × 8 prediction error block is subjected to a two-dimensional inverse DCT transform process, and decoded data 128 of the 8 × 8 prediction error block is output.

Decoded data 12 of 8 × 8 prediction error block
When 8 is input to the second 4 × 4 block divider 112, the second 4 × 4 block divider 112 outputs the decoded data 128 of the 8 × 8 prediction error block to the first 4 × 4 block. 4 obtained by dividing the frame by the divider 101
Divide into 4 × 4 pixel blocks and output decoded data 129 corresponding to 4 × 4 prediction error blocks.

The adder 109 uses the data 129 of the 4 × 4 prediction error block and the data 12 of the 4 × 4 prediction reference block.
An addition process of adding 6 and 6 is performed and decoded data 130 of a 4 × 4 pixel block is output. Here, in the addition processing, processing for adding pixel levels of corresponding pixels between the prediction error block and the prediction reference block is performed for each pixel.

Then, the decoded data 130 of the 4 × 4 pixel block output from the adder 109 is stored in the reference frame memory 110 as the data of the reference frame to be referred to in the predictive coding for the next frame of the target frame. To be done.

As described above, according to the first embodiment, motion compensation is performed in units of blocks of 4 × 4 pixels, and DC
Since the T process is performed in blocks of 8 × 8 pixels, fine motion detection can be performed, and further, the compression efficiency is reduced when the block size is reduced. It is possible to avoid a decrease in compression efficiency in the DCT process. You can In particular,
When the image size of the input image data is small, it is possible to improve the prediction efficiency by increasing the accuracy of motion detection and compress the image data more efficiently with a small bit amount.

In the first embodiment, the unit of motion compensation is a block of 4 × 4 pixels and the unit of DCT processing is a block of 8 × 8 pixels.
The unit for performing motion compensation and DCT processing is not limited to that of the first embodiment, and the block for performing DCT processing is obtained by collecting a plurality of blocks for performing motion compensation. If

For example, the unit for motion compensation is a block consisting of 2 × 8 pixels (2 × 8 pixel block), and the DCT
The processing is 8 × 8, which is a combination of these 4 2 × 8 pixel blocks.
You may perform by the block (8x8 pixel block) which consists of a pixel.

Further, although the case where the DCT processing is performed for every four units of motion compensation has been shown in the first embodiment, D
The unit of CT processing is not limited to this, but DC
The T process may be performed for each of a plurality of motion compensation units.

Furthermore, in the first embodiment, the DCT is used as the frequency conversion processing of the image data, but the frequency conversion processing of the image data is not limited to this, and any method such as Hadamard conversion may be used.

In the first embodiment, the DCT coefficient quantization process is the output 114 of the DCT unit 104.
A plurality of DCT coefficients of the × 8 prediction error block are set to each DCT.
Although the quantization processing is performed for each coefficient, the DCT coefficient quantization processing is not limited to that in the first embodiment, and may be vector quantization, for example.

Specifically, the vector quantization of the DCT coefficient is performed by a plurality of Ds corresponding to one 8 × 8 prediction error block.
The CT coefficient is represented by one vector whose components are the values of these DCT coefficients, and the vector corresponding to the 8 × 8 prediction error block is one of a plurality of vectors having components whose values are predetermined. This is a process of approximating.

Such vector quantization processing is also carried out in units of 8 × 8 prediction error blocks consisting of four 4 × 4 prediction error blocks, as in the quantization processing of the first embodiment.

Note that P116-138 (Chapter 6 Concept of Vector Quantization) in "Introduction to Advanced Technology" series published by Ohmsha (H3.8.25 supervised by Hiroshi Harashima edited by the Image Information Compression Television Society of Japan) Various vector quantization methods are described.

(Second Embodiment) FIG. 3 is a block diagram for explaining an encoding apparatus according to the second embodiment of the present invention. Like the first embodiment, the encoding device 100b according to the second embodiment performs motion compensation in units smaller than the unit for performing DCT processing when predictive encoding of image data input as a video signal. However, the coding apparatus 100b according to the second embodiment is adaptive in switching between the two motion compensation modes during the predictive coding.
This is different from the encoding device 100a of the first embodiment.

That is, the coding apparatus 100b according to the second embodiment divides the image data of each input frame according to the first embodiment into data corresponding to a 4 × 4 pixel block, which is the first 4 × 4. In addition to the block divider 101, the input image data of each frame is divided into data corresponding to a 2 × 8 pixel block in which the number of pixels in the horizontal direction is 2 and the number of pixels in the vertical direction is 8 × 2 × 8 block divider 201
have.

The coding apparatus 100b performs motion detection in units of the data 121 corresponding to the 4 × 4 pixel block, and the motion vector 12 for the 4 × 4 pixel block.
The prediction reference data (4
In addition to the motion compensating unit 111 according to the first embodiment that outputs (data of a × 4 prediction reference block) 126, motion detection is performed in units of data corresponding to a 2 × 8 pixel block, and The motion vector 224 and prediction reference data (2 × 8) corresponding to the 2 × 8 pixel block
8 prediction reference block data) 226 and the motion compensation unit 211.

The encoding apparatus 100b performs a process of subtracting the data 126 of the 4x4 prediction reference block from the data of the 4x4 pixel block which is the output of the 4x4 block divider 101, thereby performing 4x4. In addition to the subtractor 102 of the first embodiment that outputs the pixel block difference data (prediction error block data) 127, 2 from the 2 × 8 pixel block data output from the 2 × 8 block divider 201.
It has a subtracter 202 that performs a process of subtracting the data 226 of the × 8 prediction reference block and outputs difference data (data of the prediction error block) 227 of the 2 × 8 pixel block.

The encoding device 100b has the subtracter 1
02 output (4 × 4 prediction error block data) 127
And two outputs of the subtracter 202 (2 × 8 prediction error block data) 227, two motion compensation modes (4 × 4 pixel block unit motion compensation and 2 × 8 pixel block unit motion compensation mode) It has an evaluator 231 which selects one of the above and outputs a motion compensation mode identification signal (hereinafter also referred to as an evaluation signal) 240 for identifying the mode.

The selection of the motion compensation mode by the evaluator 231 is specifically performed by using the coding efficiency as a criterion. For example, the dispersion value of the pixel level in the 4 × 4 prediction error block is compared with the dispersion value of the pixel level in the 2 × 8 prediction error block, and the mode using the prediction error block with the small dispersion value is selected.

Although the pixel-level variance value in the prediction error block is used as the measure for selecting the motion compensation mode here, the measure for selecting the motion compensation mode is not limited to this. Any scale may be used as long as it is a scale corresponding to efficiency (that is, one indicating the high image quality obtained with the same code amount). High coding efficiency means that a reproduced image with higher image quality can be obtained with a small code amount.

Further, as a measure showing the coding efficiency,
Even if the sum of absolute pixel level of the prediction error block is used,
Alternatively, a distribution of transform coefficients obtained by performing transform processing such as DCT of image data may be used. For example, the plurality of DCT coefficients (frequency components) obtained by the DCT processing of image data indicate that the higher the degree of concentration on the low frequency side, the higher the coding efficiency.

As a measure for selecting the motion compensation mode, the code amount generated when variable length coding is performed on the quantized coefficient obtained by quantizing the DCT coefficient may be used. This measure uses that the coding efficiency is higher in a motion compensation mode with a smaller code amount even if the quantization step in the quantization with respect to the DCT coefficient is the same.

As a measure for selecting the motion compensation mode, the entropy of the prediction error block, that is, the pixel level variation may be used. Also, the magnitude of the motion vector may be used as a measure for selecting the motion compensation mode. This is because the smaller the motion vector of the prediction error block is, the higher the coding efficiency can be.

Also, the motion compensation mode may be determined according to the magnitude of the motion vector of the 4 × 4 pixel block located around the 4 × 4 pixel block to be processed in the frame. Since a plurality of blocks located in a moving object have almost the same motion, the motion vector of each block often has a close value.

Therefore, when the block to be processed (target block) has a motion vector having a value close to the value of the motion vector of the block (peripheral block) located in the periphery of the block, the motion detection of the block is highly accurate. Therefore, it can be determined that the coding efficiency is improved. Further, when the difference vector between the motion vector of the target block and the motion vector of the peripheral block is coded as the motion vector of the target block, it is determined that the smaller the value of the difference vector is, the higher the coding efficiency is. it can. Furthermore, as a measure of the coding efficiency, a combination of the above-mentioned plurality of measures may be used.

Further, the coding apparatus 10 of the above-mentioned second embodiment
0b is the output 127 of the subtractor 102 and the subtractor 2
It has an 8 × 8 block construction unit 232 which generates 8 × 8 prediction error block data 222 in accordance with the evaluation signal 240 from the evaluator 231 based on the output 227 of 02.

The encoding apparatus 100b performs DCT processing on the data 222 of the 8 × 8 prediction error block D
A CT unit 104, a quantizer 105 that performs a quantization process on an output 114 of the DCT unit 104 with a predetermined quantization step, and a variable length coding process on an output 123 of the quantizer 105, The coded data obtained by adding the code of the motion vectors 124 and 224 from the motion compensators 111 and 211 and the code of the evaluation signal 240 from the evaluator 231 to the coded data obtained by the variable length coding. A variable length encoder 206 that outputs the sequence Bs2.

The encoding apparatus 100b performs an inverse DCT process on the output 123 of the quantizer 105 and an inverse quantizer 107 which performs an inverse quantization process on the output 123 of the inverse quantizer 107. The inverse DCT unit 108 outputs the decoded data 128 of the 8 × 8 prediction error block.

Then, the encoding device 100b corresponds the decoded data 128 of the 8 × 8 prediction error block to the 4 × 4 prediction error block or the 2 × 8 prediction error block according to the evaluation signal 240 from the evaluator 231. And the decoded data 22 of the 4 × 4 prediction error block
Decoded data 22 of 9a or 2 × 8 prediction error block
It has a block divider 233 that outputs 9b.

The encoding device 100b has the evaluator 2
A selector 234 for selecting and outputting one of the output 126 of the motion compensating section 111 and the output 226 of the motion compensating section 211 based on the evaluation signal 240 from 31 and the selector 23.
4 output 126 or 226 and the block divider 2
33 output 229a or 229b is added and output, and 4 × which is the output of the adder 209
Decoded data corresponding to a 4-pixel block (hereinafter, also referred to as 4 × 4 decoded block data) 230a or 2 ×
A reference frame memory that stores decoded data (hereinafter, also referred to as 2 × 8 decoded block data) 230b corresponding to an 8-pixel block as data of a reference frame to be referred to in predictive coding of a frame next to a target frame. 11
It has 0 and.

The coding apparatus 10 according to the second embodiment
0b 4 × 4 block divider 101, subtractor 102, motion compensator 111, DCT device 104, quantizer 105, inverse quantizer 107, inverse DCT device 108, variable length encoder 1
Each of 06 is exactly the same as that of the first embodiment shown in FIG.

Next, the operation will be described. When the video signal is input as the image data Id to the encoding device 100b of the second embodiment, the 4 × 4 block divider 101
Divides the image data Id of the input frame so as to correspond to a 4 × 4 pixel block, outputs data 121 corresponding to a 4 × 4 pixel block, and the 2 × 8 block divider 201 inputs Frame image data Id
Is divided into 2 × 8 pixel blocks to obtain 2 ×
The data 221 corresponding to the 8-pixel block is output.

The motion compensation section 111 uses the reference frame memory 110 based on the data 121 of the 4 × 4 pixel block.
Motion detection is performed by referring to the image data of the reference frame stored in, and the motion vector 124 corresponding to the 4 × 4 pixel block to be processed and the prediction reference data (4 The data of (× 4 prediction reference block) 126 is output. Also, the motion compensation unit 211
Corresponds to the 2 × 8 pixel block to be processed by referring to the image data of the reference frame stored in the reference frame memory 110 based on the 2 × 8 pixel block data 221 to detect motion. Motion vector 22
4 and prediction reference data (data of 2 × 8 prediction reference block) 226 corresponding to the 2 × 8 pixel block is output.

The subtractor 102 uses the data 121 of the 4 × 4 pixel block to the data 126 of the 4 × 4 prediction reference block.
The subtraction process of subtracting is performed, and difference data of the 4 × 4 pixel block (data of the 4 × 4 prediction error block) 127 obtained by the subtraction process is output. Also, the subtractor 202
Performs a subtraction process of subtracting the data 226 of the 2 × 8 prediction reference block from the data 221 of the 2 × 8 pixel block, and the difference data of the 2 × 8 pixel block (2 × 8 prediction error block) obtained by the subtraction process. Data) 227 is output.

The 4 × 4 prediction error block data 127 and the 2 × 8 prediction error block data 227 are evaluated by the evaluator 231.
Then, the evaluator 231 selects one of two motion compensation modes, that is, a motion compensation mode in units of 4 × 4 pixel blocks and a motion compensation mode in units of 2 × 8 pixel blocks, based on the data of both blocks. , And outputs the motion compensation mode identification signal 240 for identifying the selected mode. Here, the method of selecting the motion compensation mode selects the coding efficiency as a criterion. For example, the variance value of the data 127 of the 4 × 4 prediction error block is compared with the variance value of the data 227 of the 2 × 8 prediction error block, and the motion compensation mode corresponding to the smaller variance value is selected.

Here, the variance value of the prediction error block is used as a measure for selecting the two motion compensation modes. However, the scale for selecting the two motion compensation modes is not limited to that of the second embodiment, and for example,
Any measure may be used as long as it corresponds to the coding efficiency as shown in the first embodiment.

The 8 × 8 block construction unit 232 uses the data 127 of the 4 × 4 prediction error block and the data 227 of the 2 × 8 prediction error block in accordance with the motion compensation mode identification signal 240 to calculate the difference of the 8 × 8 pixel block. Data (8
Data of a × 8 prediction error block) 222 is generated.

FIG. 4 shows three structural examples (FIGS. (A) to (c)) of the 8 × 8 prediction error block. For example, in FIG.
In (a), one 8 × 8 prediction error block BL8a has four 4 × 4 prediction error blocks BL4a as in the first embodiment.
~ BL4d is shown.

In FIG. 4 (b), one 8 × 8 prediction error block BLb is replaced by four 2 × 8 prediction error blocks BL28a-B28B.
It shows a case of being composed of L28d. Figure 4 (c)
Is an 8 × 8 prediction error block BL8c
× 4 prediction error blocks BL4a and BL4c and two 2 ×
It shows a case where it is composed of eight prediction error blocks BL28c and BL28d.

As described above, the data 222 of the 8 × 8 prediction error block is composed of prediction error data obtained by combining a plurality of motion compensation modes.

Data 22 of the above 8 × 8 prediction error block
2 is input to the DCT device 104, the DCT device 104
Is 2 for the data 222 of the 8 × 8 prediction error block.
A three-dimensional discrete cosine transform (DCT) process is performed, and DCT coefficient data 114 including a plurality of DCT coefficients of the 8 × 8 prediction error block is output. The quantizer 105 quantizes each DCT coefficient of the 8 × 8 prediction error block in a predetermined quantization step and outputs quantized coefficient data 123 including a plurality of quantized coefficients corresponding to the 8 × 8 prediction error block. . The plurality of quantized coefficients corresponding to the 8 × 8 prediction error block have a two-dimensional array.

The variable length encoder 206, based on the quantized coefficient data 123, quantizes a plurality of quantized coefficients having a two-dimensional array corresponding to the 8 × 8 prediction error block into a one-dimensional quantized coefficient. Then, a variable length coding process is performed in which the quantized coefficient having the one-dimensional array obtained by the scanning is coded using a variable length code. Then, the variable length encoder 206 adds the motion vector 1 obtained by the motion compensation unit 111 to the encoded data obtained by the variable length encoding process of the quantized coefficient.
The code of 24 or the code of the motion vector 224 obtained by the motion compensating unit 211 and the code of the identification signal 240 indicating the selected motion compensation mode are added and the encoded data string Bs2 is output.

The code of the motion compensation mode identification signal 240 may be arranged at the head of the code string of each prediction error block or at the head of the code string corresponding to the frame. Its sequence position in is not limited.

When the quantized coefficient data 123 of the 8 × 8 prediction error block is input to the inverse quantizer 107, the inverse quantizer 107 uses the quantization step used in the quantizer 105 to Inverse quantization processing is performed on the quantized coefficient data 123, and a plurality of D of 8 × 8 prediction error blocks
The DCT coefficient data 115 composed of CT coefficients is restored.
When the DCT coefficient data 115 of the 8 × 8 prediction error block is input to the inverse DCT device 108, the inverse DCT device 108
The DCT coefficient data 115 of the 8 × 8 prediction error block is subjected to a two-dimensional inverse DCT transform process, and decoded data 128 of the 8 × 8 prediction error block is output.

Decoded data 12 of 8 × 8 prediction error block
When 8 is input to the block divider 233, the block divider 233 outputs the decoded data 128 of the 8 × 8 prediction error block to 4 according to the motion compensation mode identification signal 240.
Decoded data 229a or 2 of a × 4 prediction error block
It is divided into decoded data 229b of a × 8 prediction error block. This block division is performed by the 8 × 8 block construction unit 232.
This corresponds to the inverse process of the 8 × 8 block generation performed in step. Decoded data 229 of the divided prediction error block
The value a or 229b is input to the adder 209.

Further, the selector 234 uses the motion compensation mode identification signal 240 to determine the motion compensation unit 1 in units of 4 × 4 pixels.
Data 126 of 4 × 4 prediction reference block obtained in 11
And 2 × obtained by the motion compensation unit 211 in units of 2 × 8 pixels
One of the data 226 of the 8 prediction reference blocks is selected and output. That is, when the decoded data output from the block divider 233 is the decoded data of the 4 × 4 pixel block, the data 12 of the 4 × 4 prediction reference block
6 is selected, on the other hand, if the decoded data output from the block divider 233 is the decoded data of the 2 × 8 pixel block, the data 226 of the 2 × 8 prediction reference block is selected.

The adder 209 adds the 4 × 4 prediction reference block data 126 output from the selector 234 and the decoded data 229a of the 4 × 4 prediction error block, or outputs 2 × output from the selector 234. The addition process of the data 226 of the 8 prediction reference block and the decoded data 229b of the 2 × 8 prediction error block is performed and the data 230a of the 4 × 4 pixel block or the decoded data 230b of the 2 × 8 pixel block is output.

The decoded data 230a of the 4 × 4 pixel block or the decoded data 230b of the 2 × 8 pixel block output from the adder 223 is the reference frame memory 1
10 and is stored as data of a predictive reference frame used in predictive encoding of the frame next to the target frame.

As described above, in the second embodiment, the area which is a unit for motion compensation is a 2 × 8 pixel block and a 4 × 4 pixel block.
Since it is adaptively switched to a predetermined one of the pixel blocks, finer motion detection can be performed, and at the same time D
Since CT processing is performed in units of 8 × 8 pixel blocks, it is possible to maintain high coding efficiency.

The motion compensation mode may be determined in 8 × 8 pixel block units or frame units, and the unit determined by the motion compensation mode is not limited.

Also, the motion compensation mode may be determined in units smaller than the 8 × 8 pixel block. An example of determining the motion compensation mode in units of 4 × 8 pixels will be specifically described with reference to FIG. Two 4 × 4 prediction error blocks B
The motion compensation mode is determined with L4a and BL4c as units, and two 2 × 8 prediction error blocks BL28c and BL28d are determined.
By determining the motion compensation mode in units of
The motion compensation mode can be determined in units of 4 × 8 pixels.

Also, the motion compensation is performed by using a plurality of types of blocks having different shapes as a unit, and the DCT process is performed by using a large block obtained by combining various blocks according to a predetermined arrangement pattern as a unit. Good. In this case, the arrangement pattern of the blocks (small blocks) as the unit of motion compensation in the large block that is the unit of DCT processing is encoded based on the encoding efficiency corresponding to each arrangement pattern obtained in advance. It is sufficient to select the arrangement pattern that maximizes the efficiency. Further, in this case, the variable length encoder 206 needs to output a code that specifies the selected block arrangement pattern. The code specifying this arrangement pattern may be arranged at the beginning of the code string corresponding to the block or at the beginning of the code string corresponding to the frame. The position of the specified code is not limited.

As described above, a plurality of types of small blocks having different shapes are used as a block which is a unit for motion compensation, and an arrangement pattern of small blocks which is a unit for performing the motion compensation in a large block which is a unit for performing DCT processing. By enabling a plurality of selections, it is possible to perform finer motion detection, and it is possible to improve coding efficiency.

4 in the 8 × 8 prediction error block
The arrangement pattern of the × 4 pixel block or the 2 × 8 pixel block is not limited to that shown in FIGS. 4 (a) to 4 (c).

The shape and size of the small block for motion compensation and the shape and size of the large block for DCT processing are not limited to those shown in the second embodiment. Any shape and size may be used as long as the DCT coding process is performed in units of a large block composed of a plurality of small blocks that are units for motion compensation.

In the above embodiment, the DCT processing is performed in units of blocks (large blocks) each including four motion compensation units (small blocks).
The number of small blocks as a unit of motion compensation processing that composes a large block as a unit of DCT processing is not limited to that shown in the above-described embodiment, and any number of small blocks may be used.

In the second embodiment, the DCT process is used as the frequency conversion process, but Hadamard transform or the like may be used for the frequency conversion process.

In the second embodiment, the quantization processing of the DCT coefficient is the output 114 of the DCT unit 104.
A plurality of DCT coefficients of the × 8 prediction error block are set to each DCT.
Although the quantization processing is performed for each coefficient, the DCT coefficient quantization processing may be the vector quantization described in the first embodiment.

Further, the coding apparatus according to the first and second embodiments has a coding process according to the conventional MPEG coding system in addition to the configuration (first coding unit) shown in FIGS. 1 and 3, respectively. A unit (second encoding unit) may be included, and the first and second encoding units may be switched according to the image size of the image data to be encoded.

For example, if the unit of motion compensation is reduced when the image size is large, the motion detection process increases. Therefore, the encoding method according to the first and second embodiments of the present invention, which reduces the unit for motion compensation, is particularly effective when the image size is small. From this fact, when the image size is larger than the predetermined size, the encoding process by the MPEG encoding method is performed as usual, and when it is not, the encoding method described in the embodiment of the present invention is used. It is possible to improve the coding efficiency while appropriately holding the data processing amount at the time of coding.

The image size that serves as a reference for switching the encoding process may be, for example, a size of about 352 × 286 pixels, but is not particularly limited to this size.

(Embodiment 3) FIG. 5 is a block diagram showing an encoding apparatus according to Embodiment 3 of the present invention. The encoding apparatus 100c according to the third embodiment performs a predictive encoding process in which a data unit for performing motion compensation on image data Id input as a video signal is smaller than a data unit for performing frequency conversion on the image data Id. One encoding unit (hereinafter also referred to as a small block motion compensation encoding unit) 1102 and a data unit for performing motion compensation on the input image data Id are larger than a data unit for performing frequency conversion on the image data Id. The first coding unit 1102 includes a second coding unit (hereinafter, also referred to as an MPEG coding unit) 1103 that performs predictive coding processing, and a resolution designating signal Is generated by a request from a user. Encoding process in the second encoding unit 1103
This is to switch between the encoding processing in (1) and (2).

Here, the first encoding unit 1102 is
The coding apparatus 100a according to the first embodiment is used. In addition, the second encoding unit (MPEG encoding unit) 1103 is configured as shown in FIG.
The coding apparatus 700 performs coding processing according to the conventional MPEG coding method shown in FIG. However, the first encoding unit 1102 may include the encoding device 100b according to the second embodiment.

The encoding device 100c performs resolution conversion processing on the input image data Id so that the image size is designated by the resolution designating signal Is, and outputs image data TId whose image size is transformed. Converter 1
100, a selector 1101 for supplying the image data TId whose image size has been converted based on the resolution designation signal to one of the small block motion compensation encoding unit 1102 and the MPEG encoding processing unit 1103, and the resolution. Upon receiving the designation signal Is, the selector 1101 uses the signal Is to send the first encoding unit 1102 and the second encoding unit 1.
Encoding identification code generation unit 110 that outputs a code (encoding identification code) Cid that identifies which of 103 is selected
4 and. The coded identification code Cid may be multiplexed in the coded data string Bs3 output from each of the coding units.

Here, the resolution converter 1100, which performs resolution conversion to reduce the image size of the image data Id, includes a low-pass filter that receives the image data Id and a pixel thinning unit that performs pixel thinning processing on the filter output. Can be achieved by A resolution converter 1100 that performs resolution conversion to increase the image size of the image data Id.
Can be realized by an interpolation filter that receives the image data as an input and a pixel interpolation unit that performs pixel interpolation processing on the filter output. Further, specifically, when the image size is smaller than the predetermined size, the selector 1101 supplies the image data Tid whose image size has been converted to the small block motion compensation encoding unit 1102, and otherwise,
The image data TId is supplied to the MPEG encoding unit 1103. The image size that serves as a reference for switching the encoding unit may be, for example, 352 × 286 pixels, but is not limited to this.

Next, the operation will be described. When the image data Id is input to the encoding device 100c according to the third embodiment, the resolution converter 1100 generates an image size of the image data Id with respect to the image data Id due to a user request or the like. The resolution conversion process is performed so that the image size indicated by the resolution designation signal Is is obtained, and the resolution-converted image data TId is output.

The selector 1101 has the resolution designating signal I
The resolution-converted image data TId based on s
It is supplied to one of the small block motion compensation coding unit 1102 and the MPEG coding processing unit 1103. Specifically, the image size indicated by the resolution designation signal Is is a reference image size for switching the encoding unit (for example, 352 × 28).
If it is larger than 6 pixels), the resolution-converted image data TId is supplied to the MPEG encoding processing unit 1103. On the other hand, the image size indicated by the resolution designation signal Is becomes the reference image size for switching the encoding unit (for example, 352).
X286 pixels) or less, the resolution-converted image data TId is the small block motion compensation encoding unit 110.
2 is supplied.

Then, the small block motion compensation coding unit 11
When the image data TId is input to the image data 02, the image data TId is input in the same manner as the encoding device 100a of the first embodiment.
Prediction coding processing is performed on the coded data string Bs
3 is output. Also, the MPEG encoding processing unit 1103
When the image data TId is input, performs predictive encoding processing on the image data TId, and outputs the encoded data string Bs3, as in the conventional encoding device 700 shown in FIG.

At this time, the coding identification code generator 1
Reference numeral 104 denotes the resolution specifying signal I in the selector 1101.
A code (encoding identification code) Cid1 for identifying which of the first and second encoding units 1102 and 1103 is selected based on s is output.

As described above, in the third embodiment, the first coding is performed in which the unit of motion compensation for the input image data Id is smaller than the unit of frequency conversion for the image data Id. The unit 1102 and the unit for performing motion compensation on the input image data Id are
A second encoding unit 1103 that performs a predictive encoding process that is larger than a unit that performs frequency conversion on the image data Id is provided, and an encoding unit that is suitable for the image size is selected. It is possible to improve the coding efficiency while making the appropriate processing amount.

Further, the image data is the first encoding unit 110.
A code (encoding identification code) C for identifying which encoding unit of the second and second encoding units 1103 has encoded.
Since the encoding identification code generation unit 1104 that outputs id is included, the decoding side can identify the encoding method for the encoded data string by the encoding identification code, and an appropriate decoding method at the time of decoding. Can be selected.

In the third embodiment, the coding device that switches the coding process according to the image size is shown. However, the switching of the coding process according to the third embodiment is performed according to the coding bit rate. You may do it.

For example, the conventional MPEG coding system coding apparatus and the coding apparatus according to the first or second embodiment of the present invention are provided, and coding processing is performed on input image data at a coding bit rate. The device may be switched. This is because the prediction efficiency in inter-frame predictive coding greatly contributes to the improvement of the coding efficiency when the coding bit rate decreases. Therefore, when the coding rate is smaller than the reference rate, the predictive coding is performed on the image data by the coding apparatus according to the first or second embodiment of the present invention, and when not, the conventional MPEG coding method coding is performed. The device may perform the predictive coding process on the image data. In this case, the reference value of the encoding bit rate for switching the encoding device may be about 1 Mbps, but it is not particularly limited to this value.

(Embodiment 4) FIG. 6 is a block diagram showing an encoding apparatus according to Embodiment 4 of the present invention. The encoding device 100d according to the fourth embodiment uses an encoding rate designation signal generated by a request from a user,
The predictive coding process in which the data unit for motion compensation is smaller than that in the DCT process and the predictive coding process in which the data unit for motion compensation is larger than that in the DCT process are switched.

That is, in the encoding apparatus 100d, the first code for performing the predictive encoding process is such that the data unit for motion compensation for the input image data Id is smaller than the data unit for frequency conversion for the image data Id. An encoding unit (small block motion compensation encoding unit) 1302, and a second unit that performs a predictive encoding process in which a data unit for performing motion compensation on the input image data Id is larger than a data unit for performing frequency conversion on the image data Id. And an encoding unit (MPEG encoding unit) 1303 of the first encoding unit 1302, and an encoding rate designation signal Ir generated by a request from a user.
The encoding process is switched to the encoding process in the encoding unit 1303. Here, the first encoding unit 1302 includes the encoding device 100a according to the first embodiment. Also, MPE
The G coding unit 1303 is composed of a coding device 700 that performs coding processing according to the conventional MPEG coding method shown in FIG. However, the first encoding unit 1302 may include the encoding device 100b according to the second embodiment. In addition, each of the above-mentioned coding units has a coding rate designation signal I
The encoding process is performed at the encoding rate according to r.

Also, the above-mentioned encoding device 100d uses the small block motion compensation encoding unit 13 to convert the image data TId whose image size has been converted in accordance with the encoding rate designation signal Ir.
02 and the MPEG encoding processing unit 1103, and the selector 1301 selects which of the first and second encoding units is selected by the selector 1301 according to the encoding rate designating signal Ir. The coded identification code generation unit 13 that outputs a code (coded identification code) Cid2 that identifies whether or not
04 and. The coding identification code Cid2 is
You may make it multiplex to the encoding data string Bs3 output from each said encoding part.

Next, the operation will be described. When the image data Id as a video signal is input to the encoding device 100d of the fourth embodiment, the selector 1301 outputs the image data Id based on the encoding rate designation signal Ir.
It is supplied to one of the small block motion compensation coding unit 1302 and the MPEG coding processing unit 1303. Specifically, when the rate indicated by the encoding rate designation signal Ir is higher than the rate (1 Mbps) that is the reference for switching the encoding section, the image data Id is the MPEG encoding section 110.
3 is supplied. On the other hand, when the rate indicated by the coding rate designation signal Ir is equal to or lower than the rate (for example, 1 Mbps) that is the reference for switching the coding unit, the image data Id
Is supplied to the small block motion compensation coding unit 1302.

Then, the small block motion compensation coding unit 13
When the image data Id is input, 02 performs the predictive coding process on the image data at the rate indicated by the coding rate designating signal Ir similarly to the coding apparatus 100a of the first embodiment, and performs coding. The data string Bs4 is output. Further, in the MPEG encoding processing unit 1303, when the image data Id is input, the predictive encoding process for the image data is performed in the same manner as the conventional encoding device 700 shown in FIG. Go in,
The encoded data string Bs4 is output.

At this time, the coding identification code generator 1
Reference numeral 304 denotes the selector 1301 for the first and second encoding units 130 based on the encoding rate designation signal Ir.
A code (coded identification code) Cid2 for identifying which of 2 and 1303 is selected is output.

As described above, in the fourth embodiment, the first coding is performed in which the unit of motion compensation for the input image data Id is smaller than the unit of frequency conversion for the image data Id. The unit 1302 and the unit for performing motion compensation on the input image data Id are
Since the second encoding unit 1303 that performs the predictive encoding process larger than the unit for performing the frequency conversion on the image data Id is provided and the encoding unit suitable for the encoding rate is selected,
It is possible to improve the coding efficiency while making the data processing amount at the time of predictive coding appropriate.

Further, the image data Id is the first encoding unit 1
The decoding side has a coding identification code generation unit 1304 that outputs a code (coding identification code) Cid2 that identifies which of the coding unit 302 or the second coding unit 1303 has coded. In, it is possible to identify the encoding method for the encoded data string by the encoding identification code, and to select an appropriate decoding method at the time of decoding.

Incidentally, the third embodiment or the fourth embodiment.
In the above, an encoding device that switches the encoding unit that encodes the image data according to the resolution of the image data or the encoding rate for the image data is shown. It may be performed according to the output system of the column. For example, as an output system of the encoded data sequence, a recording system for recording the encoded data sequence on a recording medium, a transmission system for transmitting the encoded data sequence to a transmission path, and the like are conceivable. Further, the number of recording systems or transmission systems that output the encoded data is not limited to one, and encoded data strings may be output to a plurality of types of recording systems or a plurality of types of transmission systems. .

For example, with respect to a DVD device, in order to maintain compatibility with other devices, the DVD standard stipulates that recording is performed by the processing of the MPEG encoding system. On the other hand, a recording medium such as a hard disk installed in a video recording / reproducing device has few compatibility problems and can be recorded by a free encoding method. Therefore, when outputting the encoded data string to the recording system for recording on the DVD, the conventional MPEG encoding processing is performed, and when outputting to the recording system for recording on the hard disk, the first and the second embodiments of the present invention are used. Encoding processing such as 2 may be performed.

(Fifth Embodiment) FIG. 7 is a block diagram for explaining a recording apparatus according to the fifth embodiment of the present invention. The recording apparatus 100e according to the fifth embodiment encodes image data input as a video signal, and encodes the image data from a plurality of recording media according to a media designation signal Im generated by a user's request or the like. A recording medium on which image data is to be recorded is selected, and the coded image data is recorded on the selected recording medium.

That is, the recording apparatus 100e performs the first encoding in which the data unit for motion compensation of the input image data Id is smaller than the data unit for frequency conversion of the image data Id. Unit (small block motion compensation encoding unit) 1402 and a second unit for performing predictive encoding processing in which a data unit for performing motion compensation on the input image data Id is larger than a data unit for performing frequency conversion on the image data Id. And an encoding unit (MPEG encoding unit) 1403. Here, the first encoding unit 1402 includes the encoding device 100a according to the first embodiment. The second coding unit 1403 is composed of a coding device 700 that performs coding processing by the conventional MPEG coding method shown in FIG. However, the first encoding unit 1402 may include the encoding device 100b according to the second embodiment.

The recording device 100e supplies the input image data Id to either one of the first encoding unit 1402 and the second encoding unit 1403 according to the media designation signal Im. And an encoding identification code generation for outputting a code (encoding identification code) Cid3 for identifying which of the first and second encoding units has been selected by the selector 1401 according to the media designation signal Im. Part 14
04 and.

The recording apparatus 100e is suitable for each recording medium with respect to the coded data string Bsa generated by the small block motion compensation coding unit 1402 or the coded data string Bsb generated by the MPEG coding unit 1403. Processing such as error correction processing is performed, and the encoded data string Bsa or Bsb subjected to the processing and the encoded identification code Cid
It has a recording processing unit 1405 which outputs an encoded data string Bsa4 or Bsb4 including 3 and 3. This recording processing unit 1
405 is an encoded data string Bsa and an encoded identification code Cid3
The encoded data string Bsa4 including
6 and outputs the encoded data string Bsb and the encoded identification code C.
Data output control is performed by the media designation signal Im so that the encoded data string Bsb4 including id3 is output to the DVD 1407.

Next, the operation will be described. When the image data Id as a video signal is input to the encoding device 100e of the fifth embodiment, the selector 1401 converts the image data Id into the small block motion compensation encoding unit based on the media designation signal Im. 1402 and MPEG encoding unit 1
Supply to either one of 403.

Specifically, when the media designation signal Im designates the hard disk 1406, the image data Id
Is supplied to the small block motion compensation encoding unit 1402,
When the media designation signal Im designates the DVD 1407, the image data Id is supplied to the MPEG encoding unit 1403.

Then, the small block motion compensation coding unit 14
When the image data Id is input, 02 performs predictive encoding processing on the image data Id as in the encoding device 100a of the first embodiment, and outputs the encoded data string Bsa. Also, when the image data Id is input, the MPEG encoding unit 1403 performs predictive encoding processing on the image data Id as in the conventional encoding device 700 shown in FIG. 17, and outputs an encoded data string Bsb. To do.

At this time, the coding identification code generator 1
404 outputs a code (encoding identification code) Cid3 for identifying which of the first and second encoding sections 1402 and 1403 has been selected by the selection section 1401 based on the media designation signal Ir. .

The recording processing unit 1405 has a small block motion compensation coding unit 1402 or an MPEG coding unit 1403.
The encoded data string Bsa or Bsb generated by the above is subjected to data processing such as error correction processing suitable for each recording medium based on the media designation signal Im, and the encoded data string Bsa subjected to the data processing. Alternatively, the recording data string Bs4a or Bs4b including Bsb and the encoded identification code Cid3 is output. At this time, the recording processing unit 1405
Then, based on the media designation signal Im, data output control is performed to output the recording data string Bs4a to the hard disk 1406 and output the recording data string Bs4b to the DVD 1407.

As described above, the recording apparatus 10 according to the fifth embodiment
In 0e, image data encoded by the MPEG encoding method is recorded on the DVD, so that the DVD standard is complied with and compatibility with other DVD devices can be maintained. On the other hand, since the hard disk stores the image data that has been encoded by the encoding device of the first embodiment, it is possible to encode the image data on a medium such as a hard disk with few compatibility problems by performing high-precision motion compensation on the medium. It is possible to record the coded data string obtained by the coding process with the coding efficiency.

In the fifth embodiment, the code (encoding identification code) Cid3 for identifying the type of the encoding method selected by the media designation signal (that is, the first and second encoding units) is the hard disk. 1406 or DV
Since it is recorded in D1407, the encoding method of the recorded encoded data string can be identified by reading the encoded identification code recorded on the recording medium, and an appropriate decoding can be performed at the time of decoding. It becomes possible to select the conversion method.

The coded identification code may be multiplexed and recorded in the coded data string, or an area different from the recording area of the coded data string of the recording medium, for example, management information of the record data may be recorded. It may be recorded in the area.

The processing of recording different encoded data strings on the corresponding media may be performed in parallel and simultaneously. For example, the small block motion compensation coding unit 14
The encoded data string obtained in 02 is stored in the hard disk 140.
6, the encoded data sequence obtained by the MPEG encoding unit 1403 may be recorded on the DVD 1407 at the same time.

Further, in the fifth embodiment, the hard disk and the DVD are mentioned as the recording medium, but the kind of the recording medium is not limited, and for example, CD-
R, CD-RW, DVD-RAM, DVD-R, DVD
It may be an optical disc such as -RW, DVD + R, DVD + RW, a semiconductor memory, or a magnetic recording medium, and the number of recording media may be any number as long as it is two or more.

(Sixth Embodiment) FIG. 8 is a block diagram for explaining a recording apparatus according to a sixth embodiment of the present invention. The recording apparatus 100f according to the sixth embodiment performs an encoding process on the input image data Id, and
The coding process for the image data Id to be recorded is switched according to the type of the recording medium mounted in the recording device.

That is, the recording apparatus 1 according to the sixth embodiment
00f is a first coding unit (small block motion compensation coding) that performs a predictive coding process in which a data unit that performs motion compensation on the input image data Id is smaller than a data unit that performs frequency conversion on the image data Id. Part) 150
2 and a data unit for performing motion compensation on the input image data Id is larger than a data unit for performing frequency conversion on the image data Id.
Coding unit (MPEG coding unit) 1503. Here, the first encoding unit 1502 includes the encoding device 100a according to the first embodiment. The second encoding unit 1503 is an MPEG encoding unit including an encoding device 700 that performs an encoding process by the conventional MPEG encoding method shown in FIG. However, the first encoding unit 1502 may be the encoding device 100b according to the second embodiment.

The recording apparatus 100f discriminates the type of the recording medium 1506 inserted in the recording drive, and outputs the medium discriminating signal Sd indicating the discriminating result, and the medium discriminating section 1507, and in response to the medium discriminating signal Sd. It has a selector 1501 that supplies the input image data Id to either one of the first encoding unit 1502 and the second encoding unit 1503.

Here, the media discriminating unit 1507 may discriminate the type of the medium by reading the media information recorded on the inner peripheral portion of the recording medium mounted in the recording drive, for example. However, the type of the recording medium may be determined by the intensity of the reflected light when the surface of the recording medium is irradiated with the laser light. Further, the media discriminating unit 1507 may discriminate the type of the medium based on the cartridge shape of the medium.

The recording device 100f includes the selector 150.
1 according to the media discrimination signal Sd
Coded identification code generation section 1504 that outputs a code (coded identification code) Cid4 that identifies which of the coding sections 1502 and 1503 has been selected, and the coding generated by the small block motion compensation coding section 1502. Data string Bsa
Alternatively, the encoded data sequence Bsb generated by the MPEG encoding unit 1503 is subjected to data processing such as error correction processing suitable for each recording medium, and the encoded data sequence Bsa or Bsb subjected to the data processing is obtained. Coded data string Bsa5 or Bsb5 including coded identification code Cid4
The recording processing unit 1505 that outputs

The recording processing unit 1505, for example, based on the media discrimination signal Sd, sets the image data Id to CD-
When recording on R, CD-RW, etc., the encoded data sequence Bsa output from the small block motion compensation encoding unit 1502
When the image data Id is recorded on the DVD, the MPEG encoding unit 150 is output.
The recorded coded data Bsb5 including the coded data string output from No. 3 is output.

Next, the operation will be described. In the encoding device 100f according to the fifth embodiment, the media discriminating unit 150 is included.
7 is a recording medium 1506 inserted in the recording drive.
The media discrimination signal Sd is output.

Then, the image data Id as a video signal.
Is input, the selector 1501 supplies the image data Id to one of the small block motion compensation coding unit 1502 and the MPEG coding unit 1503 based on the media discrimination signal Sd.

Specifically, when the media discrimination signal Sd is CD
When -R or CD-RW is designated, the image data Id is supplied to the small block motion compensation coding unit 1502, and when the media discrimination signal Sd designates DVD,
The image data Id is supplied to the MPEG encoding processing unit 1503.

The small block motion compensation coding unit 1502
When the image data Id is input, the predictive encoding process is performed on the image data as in the encoding device 100a according to the first embodiment, and the encoded data string Bsa is output. Also,
When the image data Id is input, the MPEG encoding unit 1503 performs predictive encoding processing on the image data as in the conventional encoding device 700 shown in FIG. 17, and outputs an encoded data string Bsb.

At this time, the coding identification code generator 1
504 outputs a code (encoding identification code) Cid4 for identifying which of the first and second encoding units 1502 and 1503 has been selected based on the media determination signal Sd.

The recording processing unit 1505 includes a small block motion compensation coding unit 1502 or an MPEG coding unit 1503.
The encoded data sequence Bsa or Bsb generated by the above is subjected to data processing such as error correction processing suitable for the recording medium based on the media determination signal Sd, and the encoded data sequence Bsa subjected to the data processing. Alternatively, the recording data string Bs5a or Bs5b including Bsb and the encoded identification code Cid4 is output.

[0189] For example, the data is recorded on a CD-R or a CD.
In the case of recording on a medium with a looser compatibility condition with other devices such as RW, the recording processing unit 1505 records recording coded data including a coded data string output from the small block motion compensation coding unit 1502. When outputting Bsa5 and recording the data on a medium having a strict requirement for compatibility with other devices such as a DVD, the MPEG encoding unit 15
The recorded coded data Bsb5 including the coded data string outputted from No. 03 is outputted.

As described above, the recording apparatus 10 according to the sixth embodiment
At 0f, image data encoded by the MPEG encoding method is recorded on the DVD, so that the DVD standard is complied with and compatibility with other DVD devices can be maintained. On the other hand, in the medium having a loose compatibility condition, that is, the CD-R or the CD-RW, the first embodiment is used.
Since the image data encoded by the encoding device is recorded, it is possible to record the encoded data string encoded with high encoding efficiency by highly accurate motion compensation.

Further, in the sixth embodiment, the code (encoding identification code) Cid4 for identifying the type of the encoding method selected by the media discrimination signal Sd is the recording medium 15
Since it is recorded on the recording medium 06, the encoding side of the recorded encoded data sequence can be identified on the data decoding side by reading the encoded identification code recorded on the recording medium. For this, an appropriate decoding method can be selected.

In the sixth embodiment, when the coded data string is recorded on the recording medium, the coded identification code may be multiplexed and recorded on the coded data string, or the code of the recording medium may be recorded. It may be recorded in an area different from the recording area of the encoded data string, for example, in an area for recording the management information of the recording data.

Further, in the sixth embodiment, the CD-R, the CD-RW and the DVD (ROM) are mentioned as the recording medium, but the kind of the recording medium is not limited and the compatibility with other devices is not limited. The recording medium whose condition regarding the compatibility is loose may be a semiconductor memory or a magnetic recording medium, and the medium whose condition regarding the compatibility with other devices is strict is DVD-RAM, DVD-R, DVD-.
It may be an optical disc such as RW, DVD + R, DVD + RW.

(Embodiment 7) FIG. 9 is a block diagram for explaining a decoding apparatus according to Embodiment 7 of the present invention. The decoding device 100g of the seventh embodiment is for decoding the coded data string Bs1 output from the coding device 100a of the first embodiment.

That is, the decoding device 100g classifies the quantized coefficient code string 322 and the motion vector code string included in the input coded data string Bs1 into the quantized coefficient code string 322 and the motion vector. A code classifier 301 that outputs a vector 330, and a code string 322 of the quantized coefficient
A variable length decoder 302 that restores the quantized coefficient data 323 of the 8 × 8 prediction error block by performing a variable length decoding process on the image data of the decoded frame stored in the reference frame memory 308. Referring to the data 318, the image data of the area indicated by the motion vector 330 in the frame is set to the data 331 of the 4 × 4 prediction reference block.
And a motion compensation unit 307 for outputting as.

The decoding apparatus 100g performs the inverse quantization process on the quantized coefficient data 323 of the 8 × 8 prediction error block by the quantization step used at the time of encoding to obtain the 8 × 8 prediction. DCT coefficient data 31 of error block
The inverse quantizer 303 that restores 4 and the inverse two-dimensional discrete cosine transform process are performed on the DCT coefficient data 314 of the 8 × 8 prediction error block to generate the decoded data 324 of the 8 × 8 prediction error block. It has a DCT unit 304.

The decoding device 100g differs from the 16 × 16 block construction unit 804 in the conventional decoding device 800 shown in FIG. 18 in that the decoded data 324 of the 8 × 8 prediction error block is converted into four 4 × 4 prediction errors. It has a 4 × 4 block divider 305 that divides the block into decoded data 325.

The decoding apparatus 100g adds the decoded data 325 of the 4 × 4 prediction error block and the data 331 of the 4 × 4 prediction reference block to obtain the decoded data (4 × 4 pixel block) of the 4 × 4 pixel block. Data of decoding block) 34
The adder 306 that outputs 0 and the image configurator 309 that reproduces the image data corresponding to the frame from the data 340 of the 4 × 4 decoding block and that outputs the reproduced image data RId and the 4 × 4 decoding block It has a reference frame memory 308 that stores the data 340 as data to be referred to when performing predictive decoding for the next frame of the target frame.

Next, the operation will be described. When the encoded data string Bs1 from the encoding device 100a is input to the decoding device 100g, the code classifier 301 causes the code string 32 of the quantized coefficients included in the encoded data string Bs1.
2 and the motion vector code string are classified, and the code classifier 301 outputs a quantized coefficient code string 322 and a motion vector 330.

The variable length decoder 302 performs a variable length decoding process on the code string 322 of the quantized coefficient to obtain 8 × 8.
The quantized data 323 including the quantized coefficients having the two-dimensional array of the prediction error block is generated. Further, the dequantizer 303 dequantizes the quantized data 323 of the 8 × 8 prediction error block by dequantizing the quantized data 323 using the quantization step in the coding process to obtain the two-dimensional data of the 8 × 8 prediction error block. It is converted into DCT coefficient data 314 composed of DCT coefficients having an array. The inverse DCT unit 309 converts the DCT coefficient data 314 of this 8 × 8 prediction error block into
It is converted into decoded data 324 of an 8 × 8 prediction error block by inverse two-dimensional discrete cosine transform processing.

The 4 × 4 block divider 305 uses the above 8 ×
Decoded data 324 of 8 prediction error blocks is divided into four 4 ×
It is divided into decoded data 325 of four prediction error blocks. As shown in FIG. 2A, one 8 × 8 prediction error block BL8 is divided into four 4 × 4 prediction error blocks BL4a-
It corresponds to BL4d.

Also, the motion compensation unit 307 refers to the image data 318 of the decoded frame stored in the reference frame memory 308 to perform motion compensation on the image data of the area indicated by the motion vector 330 in the frame by 4 ×. The data is output as data 331 of the 4-prediction reference block.

The adder 306 adds the data 325 of the 4 × 4 prediction error block and the data 331 of the prediction reference block obtained by the motion compensation unit 307, and outputs the decoded data 340 of the 4 × 4 pixel block. Image composer 3
09 is the decoded data 34 corresponding to the 4 × 4 pixel block
The decoded data corresponding to the frame is reproduced from 0, and the reproduced image data RId is output. Further, the decoded data 340 of the 4 × 4 pixel block is accumulated in the reference frame memory 308 as the data of the predictive reference frame used in the predictive decoding of the frame next to the target frame.

As described above, according to the seventh embodiment, when the predictive decoding process is performed on the encoded data string, motion compensation is performed by 4
Inverse DCT is performed by a block composed of × 4 pixels as a unit.
Since the processing is performed in units of blocks of 8 × 8 pixels, it is possible to perform finer motion compensation and appropriately decode the image data that has been subjected to the encoding processing, and thus encoding with high encoding efficiency can be performed. A decoding process corresponding to the process can be realized.

In the seventh embodiment, the unit for motion compensation is a block consisting of 4 × 4 pixels, and the inverse DCT is performed.
Although the unit of processing is a block composed of 8 × 8 pixels, the shape of the block which is the unit of motion compensation and inverse DCT processing is not limited to that of the seventh embodiment, and inverse DCT processing is performed. The block that is a unit to be performed may be any block that can be obtained by collecting a plurality of blocks that is a unit to perform motion compensation. For example, the unit for motion compensation is 2 ×
A block consisting of 8 pixels (2 × 8 pixel block),
The inverse DCT process may be performed for each block (8 × 8 pixel block) composed of 8 × 8 pixels obtained by combining these four 2 × 8 pixel blocks.

Further, although the case where the inverse DCT processing is performed for every four units of motion compensation has been shown in the above-mentioned seventh embodiment,
The inverse DCT processing is not limited to this, and may be performed for each of a plurality of motion compensation units.

Further, in the seventh embodiment, the inverse DCT is used as the inverse frequency conversion processing of the image data, but the inverse frequency conversion processing of the image data is not limited to this, and for example, inverse Hadamard conversion or the like. But it is okay.

Further, the inverse quantization processing for converting the quantized coefficient into the DCT coefficient is not limited to that of the seventh embodiment, and for example, the quantized coefficient is obtained by vector quantization of the DCT coefficient. In some cases, it goes without saying that inverse vector quantization is used for the inverse quantization process on the quantized coefficient.

(Embodiment 8) FIG. 10 is a block diagram for explaining a decoding apparatus according to Embodiment 8 of the present invention. The decoding apparatus 100h according to the eighth embodiment is similar to the decoding apparatus 100g according to the seventh embodiment in that when performing predictive decoding on encoded image data, motion compensation is performed by inverse D
Although it is performed in a data unit smaller than the data unit in which the CT processing is performed, the decoding device 100h according to the eighth embodiment.
Is different from the decoding device 100h of the seventh embodiment in that the two motion compensation modes are adaptively switched during the predictive decoding.

That is, the decoding apparatus 1 according to the eighth embodiment
00h is the encoding device 100b according to the second embodiment of the present invention.
It is possible to decode the encoded data Bs2 generated by. 10, the same symbols as those in FIG. 9 indicate the same components as those in the decoding device 100g of the seventh embodiment.

That is, the decoding apparatus 100h according to the eighth embodiment includes the quantized coefficient code string 322, the motion vector code, which are included in the input coded data string Bs2.
And a code classifier 401 that classifies the code 409 of the motion compensation mode identification signal and outputs the code string 322 of the quantized coefficient, the motion vector 330, and the code 409 of the motion compensation mode identification signal, and the code of the motion compensation mode identification signal. Decode 409 to partition one 8x8 prediction error block, 4
The prediction block pattern decoder 402 outputs a motion compensation mode identification signal 410 for identifying the arrangement pattern of the × 4 prediction error block and the 2 × 8 prediction error block.

The decoding apparatus 100h performs variable length decoding processing on the coded string 322 of the quantized coefficients,
A variable length decoder 302 for restoring the quantized coefficient data 323 of the 8 × 8 prediction error block, and the decoded frame stored in the reference frame memory 308 based on the motion vector 330 and the motion compensation mode identification signal 410. Motion compensation is performed by referring to the image data 318 of the image data 318, and the image data in the area indicated by the motion vector 330 in the frame is output as the data 431a of the 4 × 4 prediction reference block or the data 431b of the 2 × 8 prediction reference block. Section 404.

The decoding apparatus 100h performs the inverse quantization process on the quantized coefficient data 323 of the 8 × 8 prediction error block by the quantization step used at the time of encoding to obtain the 8 × 8 prediction. DCT coefficient data 31 of error block
The inverse quantizer 303 that restores 4 and the inverse two-dimensional discrete cosine transform process are performed on the DCT coefficient data 314 of the 8 × 8 prediction error block to generate the decoded data 324 of the 8 × 8 prediction error block. It has a DCT unit 304.

Then, the decoding device 1 according to the eighth embodiment
00h is 8 × 8 according to the motion compensation mode identification signal 410 corresponding to the 8 × 8 prediction error block to be processed.
8 prediction error block data 324 is converted to 4 × 4 prediction error block data 425a or 2 × 8 prediction error block data 42 that constitutes the 8 × 8 prediction error block.
It has a block divider 403 for dividing and outputting to 5b.

Furthermore, the decoding device 1 according to the eighth embodiment
00h is an addition process of adding the data 425a of the 4 × 4 prediction error block and the data 431a of the 4 × 4 prediction reference block and outputting the decoded data 440a of the 4 × 4 prediction block, or the addition process of the 2 × 8 prediction error block. Data 42
5b and the data 431b of the 2 × 8 prediction reference block are added, and an adder 405 that performs addition processing for outputting decoded data 440b of the 2 × 8 prediction block is included.

The decoding device 100h has the adder 40
The decoded data 440 of the 4 × 4 prediction block which is the output of 5
an image configurator 408 that reproduces image data corresponding to a frame from the decoded data 440b of the a or 2 × 8 prediction block, and outputs the reproduced image data RId;
The decoded data 440a of the 4 × 4 prediction block or the decoded data 440b of the 2 × 8 prediction block, which is the output of the adder 405, is stored as the data of the reference frame to be referred to in the predictive decoding of the frame next to the target frame. And a reference frame memory 308 for performing the same.

Next, the operation will be described. When the coded data string Bs2 from the coding device 100b is input to the decoding device 100h, the code classifier 401 causes the code string 322 of the quantization coefficient included in the input coded data string Bs2. , The code of the motion vector and the code 409 of the motion compensation mode identification signal are classified, and the code string 322 of the quantized coefficient, the motion vector 330, and the code 409 of the motion compensation mode identification signal are output.

The quantized coefficient code string 322 is processed by the variable length decoder 302, the inverse quantizer 303, and the inverse DCT unit 304, as in the case of the seventh embodiment of the present invention. The decoded data 324 of the prediction error block is obtained. In addition, the data 324 of the 8 × 8 prediction error block
Is input to the block divider 403, and the code 409 of the motion compensation mode identification signal is input to the prediction block pattern decoder 402.

The prediction block pattern decoder 402 decodes the code 409 of the motion compensation mode identification signal to obtain the arrangement pattern of the 4 × 4 prediction error block or the 2 × 8 prediction error block in one 8 × 8 prediction error block. A motion compensation mode identification signal 410 for identification is output.

The block divider 403 converts the decoded data 324 of the 8 × 8 prediction error block into the 8 × 8 prediction error block according to the motion compensation mode identification signal 410 corresponding to the 8 × 8 prediction error block to be processed. Of the decoded data 425a of the 4 × 4 prediction error block or the decoded data 425b of the 2 × 8 prediction error block, which are output.

Here, the arrangement pattern of 4 × 4 prediction error blocks or 2 × 8 prediction error blocks in one 8 × 8 prediction error block is, for example, the arrangement pattern shown in FIGS. 4 (a) to 4 (c). , One of them is specified by the motion compensation mode identification signal 410.

The motion compensating section 404 performs motion compensation based on the motion compensation mode identification signal 410 and the motion vector 330, and the data of the area indicated by the motion vector 330 in the reference frame, that is, the data of the 4 × 4 prediction reference block. 431a or 2 × 8 prediction reference block data 431b is output. At this time, whether the prediction reference block is the 4 × 4 prediction reference block or the 2 × 8 prediction reference block is determined according to the motion compensation mode identification signal 410.

The adder 405 uses the data 425a of the 4 × 4 prediction error block and the data 4 of the 4 × 4 prediction reference block.
31a is added, or addition processing for adding the data 425b of the 2 × 8 prediction error block and the data 431b of the 2 × 8 prediction reference block is performed to obtain 4 × 4.
The decoded data of the prediction block (data of the 4 × 4 decoded block) 440a or the decoded data of the 2 × 8 prediction block (data of the 2 × 8 decoded block) 440b is output.

Then, the image configurator 408 reproduces the image data corresponding to the frame from the data 440a of the 4 × 4 decoded block or the data 440b of the 2 × 8 decoded block, and outputs the reproduced image data RId. .
In addition, the data 440a or 2 of the 4 × 4 decoded block
The data 440b of the × 8 decoded block is stored in the reference frame memory 308 as the data of the reference frame used in the predictive decoding of the frame next to the target frame.

As described above, the decoding device 1 according to the eighth embodiment
At 00h, the 4 × 4 prediction error block or the 2 × 8 prediction error block in the unit of the inverse DCT process
Since the data 431a of the 4 × 4 prediction reference block or the data 431b of the 2 × 8 prediction reference block is generated according to the signal 410 indicating the arrangement pattern of the 8 prediction error blocks, as a block which is a data unit for motion compensation, It becomes possible to appropriately decode image data that has been encoded by performing more detailed motion compensation using multiple types of blocks of different shapes, and predictive decoding corresponding to predictive encoding processing with high encoding efficiency. Processing can be realized.

In the eighth embodiment, the data units for motion compensation are 4 × 4 pixel blocks and 2 × 8 pixel blocks, and the data units for inverse DCT processing are 8 ×.
Although the 8-pixel block is used, the shape of the block as the unit of motion compensation and the shape of the block as the unit of inverse DCT processing are not limited to those in the above-described eighth embodiment, and the unit of inverse DCT processing is The inverse DCT process may be performed for each block including a plurality of motion compensation blocks.

Further, although the case where the inverse DCT processing is performed for every four motion compensation data units has been shown in the above-mentioned Embodiment 8, the unit of the inverse DCT processing is not limited to this. It may be composed of a plurality of data units for motion compensation.

In the eighth embodiment, the inverse DCT is used as the inverse frequency transform for the coefficient data obtained by the frequency transform. However, the inverse frequency transform is not limited to this, and for example, the inverse Hadamard transform is possible. And so on.

In addition, the inverse quantization processing for converting the quantized coefficient into the DCT coefficient in the above-mentioned Embodiment 8 is performed, for example, by
Needless to say, when the quantized coefficient is obtained by vector quantization of the DCT coefficient, inverse vector quantization is used for the inverse quantization process on the quantized coefficient.

Furthermore, the decoding devices of the above-mentioned seventh and eighth embodiments perform decoding processing by the conventional MPEG decoding system in addition to the configuration (first decoding section) shown in FIGS. 9 and 10, respectively. A decoding unit (second decoding unit) for performing,
You may make it switch the decoding part which decodes the input coded data according to the management information of coded data.

In this case, the management information includes identification information (encoding information) indicating the encoding method used when the encoded data is generated. For example, the management information of the encoded data is recorded in a predetermined area of the recording medium, and is added to the encoded data string as its header information. With such management information, the decoding device can identify the encoding method used when generating the input encoded data string, and the decoding process suitable for the input encoded data string can be performed. It can be performed.

(Embodiment 9) FIG. 11 is a block diagram showing a decoding apparatus according to Embodiment 9 of the present invention. The decoding apparatus 100i according to the ninth embodiment includes a first decoding unit (small block motion compensation decoding unit) 1202 and a second decoding unit (MPEG decoding unit) 1203 that perform different decoding processes. Having, the encoding device 1 according to the fourth embodiment
The encoded data string Bs4 output from 00d is decoded by the decoding unit selected based on the encoded identification code Cid2 from the encoding device 100d.

Here, the first decoding unit 1202 is
In the decoding device 100g of the seventh embodiment, the data unit (4 × 4 pixel block) for motion compensation is the data unit (8 × 8 pixels) for performing the inverse DCT process on the input encoded data string Bs4. The block is smaller than the block). The first decoding unit 12
02 may be the decoding device 100h of the eighth embodiment. Also, the second decoding unit 1203
18 is a conventional MPEG decoding system decoding device 800 shown in FIG. 18, in which a data unit (16 × 16 pixel block) for motion compensation is a data unit for performing inverse DCT processing on the encoded data string Bs4. (8 × 8 pixel block)
It performs a larger predictive decoding process.

The decoding device 100i is the same as the encoding device 1.
The coded identification code decoding unit 1200 which decodes the coded identification code Cid2 from 00d and outputs the identification information Did2.
And the encoded data string Bs4 based on the identification information Did2, the first and second decoding units 1202 and 12
And a selector 1201 for supplying the signal to one of the terminals 03.

Next, the operation will be described. The decoding device 100i according to the ninth embodiment is the same as the encoding device 100d.
From the encoded identification code C
When input together with id2, the coding identification code decoding unit 1
Reference numeral 200 decodes the encoded identification code Cid2 and outputs identification information Did2, and the selector 1201 outputs the identification information Di.
Based on d2, it is determined which of the decoding units 1202 and 1203 of the first and second decoding units to decode the encoded data string Bs4, and the first and second decoding units The encoded data string Bs4 is supplied to the determined decoding unit of the units 1202 and 1203.

When the encoded data string Bs4 is input, the first decoding unit (small block motion compensation decoding unit) 1202 receives the encoded data similar to the decoding device 100g of the seventh embodiment. The column Bs4 is subjected to predictive decoding processing in which the data unit for motion compensation is smaller than the data unit for inverse DCT processing, and the reproduced data RId is output.
Further, the MPEG decoding unit 1203 uses the encoded data string B
When s4 is input, similarly to the conventional decoding device 800, the prediction decoding process in which the data unit for motion compensation is larger than the data unit for the inverse DCT process is performed, and the reproduction data RId is output.

As described above, the decoding device 1 according to the ninth embodiment
In 00i, the input encoded data string is subjected to the decoding process corresponding to the input encoded data string, based on the identification information indicating the encoding method used when the data is generated. Appropriate decoding processing can be performed on the coded data string obtained by the coding processing according to the coding rate or the like.

In the ninth embodiment, the decoding device corresponding to the coding device that switches the coding process according to the coding rate, that is, the decoding process according to the coding process selected on the coding side. However, the decoding device may switch the decoding process according to the input system of the encoded data string.

For example, as an input system for encoded data,
There are a data recording system that reads and outputs encoded data from one or a plurality of types of recording media, and a data transmission system that transmits encoded data through a transmission line having one or a plurality of types of transmission bands. Specifically, when reproducing the encoded data read from the DVD, the conventional MPEG decoding process is performed on the encoded data, and when reproducing the encoded data read from the hard disk. As with the decoding device according to the seventh or eighth embodiment of the present invention, the encoded data is subjected to predictive decoding processing in which the data unit for motion compensation is smaller than the data unit for inverse DCT processing. do it.

(Embodiment 10) FIG. 12 is a block diagram illustrating a reproducing apparatus according to Embodiment 10 of the present invention. The reproducing apparatus 100j according to the tenth embodiment is a DVD
The encoded data string Bs4b read from 1607 is decoded by the conventional MPEG decoding process, and the hard disk 1
The encoded data string Bs4a read from 606 is decoded by the predictive decoding process in which the data unit of motion compensation is smaller than the data unit of the inverse DCT process, similar to the process in the decoding device 100g of the seventh embodiment. The encoded data Bs4 output from the recording media 1406 and 1407 of the recording apparatus 100e according to the fifth embodiment.
It can reproduce a and Bs4b.

That is, the reproducing apparatus 10 of the tenth embodiment
0j is a media designation signal Im for the encoded data sequence Bs4a read from the hard disk 1606 and the encoded data sequence Bs4b read from the DVD 1607.
On the basis of the above, there is a reproduction processing unit 1605 which performs reproduction processing such as error correction processing suitable for each recording medium and outputs it. The reproduction processing unit 1605 uses the encoded data string Bs4a that has been subjected to the reproduction processing as the encoded data string Bsa.
The encoded data string Bs4b that has been subjected to the reproduction process is output as the encoded data string Bsb.

The reproducing apparatus 100j selects the encoded data string Bsa or Bsb according to the media designation signal Im, and outputs the encoded data string Bsa to the first decoding unit 1602.
And a selector 1601 for supplying the encoded data Bsb to the second decoding unit 1603.

The reproducing apparatus 100j comprises the decoding apparatus 100g of the seventh embodiment, and the encoded data sequence Bsa
A data unit (4 × 4 pixel block) for which motion compensation is performed is smaller than a data unit (8 × 8 pixel block) for which inverse DCT processing is performed, and reproduced image data RId is output. 1 decoding unit (small block motion compensation decoding unit) 1602 and the conventional MP shown in FIG.
The decoding unit 800 of the EG decoding system is used, and a data unit for performing motion compensation (16 × 16 pixel block) is a data unit for performing inverse DCT processing (8 × 8 pixel block) on the encoded data string Bsb. The second decoding unit (MPEG decoding unit) 1603 which performs a larger predictive decoding process and outputs the reproduced image data RId.

The first decoding section 1602 may be composed of the decoding device 100h of the eighth embodiment.

Next, the operation will be described. When the encoded data Bs4b read from the DVD or the encoded data Bs4a read from the hard disk is input to the reproducing device 100j of the tenth embodiment, the reproduction processing unit 1605 causes the reproduction processing unit 1605 to have the encoded data Bs4a. And Bs4b are subjected to reproduction processing such as error correction processing suitable for each recording medium based on the media designation signal Im, and the encoded data string B from the hard disk subjected to the reproduction processing.
The encoded data string Bsb from sa or the DVD subjected to the reproduction processing is output to the selector 1601.

The selector 1601 uses the encoded data string B
sa or Bsb is selected according to the media designation signal Im, and the selected encoded data string Bsa is selected by the first decoding unit 16
02 to the selected encoded data string Bsb
Is supplied to the second decoding unit 1603.

The first decoding unit 1602 performs a predictive decoding process on the input coded data sequence Bsa in which the data unit for motion compensation is smaller than the data unit for the inverse DCT process, and the reproduced image is reproduced. The data RId is output. Also, the second decoding unit 1603 performs a predictive decoding process on the input encoded data string Bsb in which the data unit for motion compensation is larger than the data unit for the inverse DCT process to reproduce image data. Output RId.

As described above, the reproducing apparatus 1 according to the tenth embodiment
At 00j, the encoded data string Bs4a read from the hard disk has the inverse DCT as the data unit for motion compensation.
A coded data string Bs4 read out from the DVD after being decoded by a predictive decoding process smaller than the data unit to be processed
Since b is decoded by the predictive decoding process in which the data unit of motion compensation is larger than the data unit of the inverse DCT process, it is recorded on a recording medium such as a hard disk with a low requirement for compatibility between devices, and is highly accurate. DVD with stringent requirements regarding compatibility between encoded data strings obtained by encoding with high encoding efficiency using motion compensation and equipment
It is possible to decode the coded data obtained by the general-purpose predictive coding process, which is recorded in the recording medium such as the above, by the predictive decoding process suitable for each.

(Embodiment 11) FIG. 13 is a block diagram for explaining a reproducing apparatus according to Embodiment 11 of the present invention. The reproducing apparatus 100k according to the eleventh embodiment switches the decoding process for the encoded data sequence according to the type of the recording medium on which the encoded data sequence is recorded. The encoded data string Bs5a or Bs5b read from the 100 f recording medium can be reproduced.

Here, the encoded data string Bs5a is a CD-R or CD whose conditions regarding compatibility with other devices are loose.
The first encoding unit (small block motion compensation encoding unit) 1 in the recording device 100f according to the sixth embodiment, which is data recorded on a recording medium such as RW, and which is image data.
It is obtained by encoding by 502. Also, the encoded data string Bs5b is data recorded on a recording medium such as a DVD, which has a strict condition regarding compatibility with other devices, and the image data is recorded in the recording device 1 of the sixth embodiment.
Second encoding unit (MPEG encoding unit) 1 in 00f
It is obtained by encoding by 503.

That is, the reproducing apparatus 10 of the eleventh embodiment
0k indicates a recording drive 1706a for accessing data to the recording medium 1706 and the recording drive 1706a.
A media discriminating unit 1707 that discriminates the type of the recording medium 1706 attached to the recording medium 1706 and outputs a medium discriminating signal Sd indicating the discrimination result, and an error correction process suitable for each recording medium according to the medium discriminating signal Sd. The encoded data string Bs5a or Bs5b which has been subjected to the reproducing process is converted into the encoded data string Bsa or Bs.
It has a reproduction processing unit 1705 which outputs as sb.

Here, the encoded data string Bs5a is the CD-
The coded data string read from R or CD-RW, and the coded data string Bs5b is the coded data string read from DVD.

The reproducing apparatus 100k selects the encoded data Bsa or Bsb according to the media discrimination designation Sd, and transmits the encoded data sequence Bsa to the first decoding section 1702.
The selector 1701 for supplying the encoded data Bsb to the second decoding unit 1703 is included.

The reproducing apparatus 100k comprises the decoding method 100g according to the seventh embodiment, and the encoded data string Bsa.
, A first decoding unit (small block motion compensation decoding unit) that performs prediction decoding processing in which the data unit for motion compensation is smaller than the data unit for inverse DCT processing and outputs reproduced image data RId 1702 and a conventional MPEG decoding method decoding apparatus 800 shown in FIG. 18, the prediction decoding in which the data unit for motion compensation is larger than the data unit for inverse DCT processing with respect to the encoded data string Bsb. Second for performing processing and outputting reproduced image data RId
And a decoding unit (MPEG decoding unit) 1703. The first decoding unit 1702 may be the decoding device 100h of the eighth embodiment.

Next, the operation will be described. In the reproducing apparatus 100k according to the eleventh embodiment, the recording drive 1706
When the recording medium 1706 is inserted into a, the medium discriminating unit 1707 discriminates the type of the recording medium and outputs a medium discriminating signal Sd indicating the discriminating result.

Then, the reproduction processing unit 1705 corrects the error of the encoded data string Bs5a or Bs5b recorded on the recording medium 1706, which is suitable for the type of the recording medium 1706, in accordance with the medium discrimination signal Sd. A reproduction process such as a process is performed, and the encoded data string Bs5a or Bs5b subjected to the process is output to the selector 1701 as an encoded data string Bsa or Bsb.

The selector 1701 supplies the input encoded data sequence Bsa to the first encoding unit 1702 and the encoded data sequence Bsb to the second encoding unit 1703 according to the media discrimination signal Sd. To do.

The first decoding unit 1702 performs predictive decoding processing on the input coded data sequence Bsa in which the data unit for motion compensation is smaller than the data unit for inverse DCT processing, The reproduced image data RId is output. The second decoding unit 1703 performs a predictive decoding process on the input coded data sequence Bsb for which the data unit for motion compensation is larger than the data unit for the inverse DCT process, and reproduces the reproduced image. The data RId is output.

As described above, the reproducing apparatus 1 according to the eleventh embodiment.
In 00k, since the decoding process for the encoded data is switched according to the type of recording medium on which the encoded data is recorded, the image data encoded by the encoding method according to the type of recording medium is changed to It can be decrypted by an appropriate decryption process.

(Embodiment 12) As Embodiment 12 of the present invention, a mobile phone equipped with the decoding apparatus of Embodiment 7 will be described below. FIG. 14 is a diagram for explaining the mobile phone according to the twelfth embodiment.

Mobile phone 1800 according to the twelfth embodiment
Outputs a radio signal N received by the antenna 1801 to the signal processing unit 1802 as a reception signal, and the signal processing unit 1802 that performs various signal processing.
The wireless communication unit 1803 transmits the transmission signal generated in 802 as the wireless signal N from the antenna 1801.

The mobile phone 1800 has a liquid crystal panel (LCD) 1806 for displaying an image, a microphone 1808 for inputting voice, a speaker 1807 for reproducing a voice signal, and an image signal for capturing an image. The camera unit 1809 to output and the image signal from the camera unit 1809 are output to the signal processing unit 1802, and the liquid crystal display unit (LCD) 1806 displays an image based on the image signal processed by the signal processing unit 1802. And an image input / output unit 1804 for controlling so that the audio input / output unit 1804 outputs the audio signal input from the microphone 1808 to the signal processing unit 1802 and outputs the audio signal processed by the signal processing unit 1802 to the speaker 1807. And a portion 1805. Note that the button operation unit of the mobile phone is not shown here for simplification of description.

Here, the signal processing section 1802 has an image decoding section (not shown) which performs the same decoding processing as the decoding apparatus of the seventh embodiment.

In the mobile phone 1800 according to the twelfth embodiment, when the code string of the image data included in the received wireless signal is decoded by the predictive decoding process, the motion compensation is 4 ×.
Since the block consisting of 4 pixels is performed as a unit and the inverse DCT processing is performed as a unit of a block consisting of 8 × 8 pixels, a fine motion compensation is performed and encoding processing is performed so as to be suitable for a mobile phone with a small image size to be handled. The image data thus obtained can be appropriately decoded.

In the twelfth embodiment, the signal processing section of the mobile phone has the decoding device of the seventh embodiment, but the mobile phone is the same as that of the first embodiment.
4 to the decoding device according to the fifth or sixth embodiment, or the reproducing device according to the ninth to eleventh embodiments.

Further, in the above-described first to eleventh embodiments, the encoding device, the decoding device, the recording device, or the reproducing device is realized by hardware, but these devices are realized by software. Good. in this case,
By recording the program for performing the data processing shown in each of the above embodiments in a data storage medium such as a flexible disk, the data processing can be constructed in an independent computer system.

In this case, the processing by the circuit elements other than the recording medium, which compose the coding apparatus, the decoding apparatus, the recording apparatus, or the reproducing apparatus of the above-described first to eleventh embodiments, may be executed by software.

FIG. 15 is a diagram for explaining a recording medium storing a program for performing the data processing of the above-described first to eleventh embodiments by software, and a computer system including the recording medium. FIG. 15A shows the appearance of the flexible disk as viewed from the front, the cross-sectional structure, and the flexible disk body, and FIG. 15B shows an example of the physical format of the flexible disk body.

The flexible disk FD has a structure in which the flexible disk body D is housed in a flexible disk case FC, and a plurality of flexible disk bodies D are concentrically formed from the outer circumference to the inner circumference. Tracks Tr are formed, and each track T
r is divided into 16 sectors Se in the circumferential direction. Therefore, in the flexible disk FD storing the program, the data as the program is recorded in the area (sector) Se allocated on the flexible disk body D.

FIG. 15C shows a configuration for recording the above program on the flexible disk FD and a configuration for performing data processing by software using the program stored in the flexible disk FD. .

The above program is executed on the flexible disk F.
When recording to D, the computer system Cs writes the data as the above program to the flexible disk FD through the flexible disk drive FDD. Further, when the devices of the above-mentioned first to eleventh embodiments are constructed in the computer system Cs by the program recorded on the flexible disk FD, the program is read from the flexible disk FD by the flexible disk drive FDD, and the computer system C is read.
load into s.

In the above description, the flexible disk is shown as the data recording medium, but an optical disk may be used as the data recording medium, and in this case, data processing by software can be performed as in the case of the flexible disk. it can. Further, the data recording medium is not limited to the optical disc and the flexible disc, and may be any one such as an IC card and a ROM cassette as long as the program can be recorded. Even when these data recording media are used, Data processing by software can be carried out as in the case of using the flexible disk or the like.

[0273]

As described above, according to the present invention (Claim 1), there is provided an encoding device for predictively encoding image data frame by frame by using pixel value correlation between frames, and a frame to be processed is Prediction processing to predict the corresponding image data,
A data predicting unit that generates predicted image data by performing each prediction block including a plurality of pixels that partition the frame;
A data encoding unit that encodes the image data of the processing target frame for each encoding block including a plurality of prediction blocks using the prediction image data corresponding to the encoding block. Therefore, it is possible to perform fine motion detection and improve coding efficiency.

That is, by increasing the block size in transform coding such as DCT, the compression efficiency by predictive coding is improved, and at the same time, the unit of motion compensation is reduced, so that finer motion detection can be performed. . As a result, the image quality of the reproduced image can be improved for a limited code amount, and particularly when the image size is small, the image quality can be greatly improved.

[0275] According to the present invention (claim 2), in the encoding device according to claim 1, the data prediction section is data for generating predicted image data corresponding to each of a plurality of types of prediction blocks having different shapes. Of the prediction image data corresponding to the plurality of types of prediction blocks having different shapes, the prediction image data of the prediction block having the highest coding efficiency with respect to the coding block is selected, and the coding is performed. Since the prediction image data corresponding to the block is generated, it is possible to perform highly accurate motion detection that matches the motion of the image, and it is possible to improve encoding efficiency.

[0276] According to the present invention (claim 3), in the encoding device according to claim 2, the data predicting unit is configured to select which shape of the predicted image data corresponding to a plurality of types of prediction blocks having different shapes. Since it is characterized in that block identification information indicating whether the predicted image data of the prediction block has been selected is output, it is possible to recognize the shape of the prediction block at the time of decoding.

According to the present invention (claim 4), in the encoding device according to claim 1, the data predicting section is data for generating predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes. An arrangement pattern that includes a generation unit and that generates prediction image data corresponding to a plurality of types of prediction blocks having different shapes so that the arrangement pattern of the various prediction blocks in the encoding block is the highest in the encoding efficiency for the encoding block. Since it is characterized in that the predicted image data corresponding to the above-mentioned coded block is generated in combination, it is possible to perform highly accurate motion detection in accordance with the shape and motion of a moving image within a frame. As a result, the coding efficiency can be improved.

According to the present invention (Claim 5), in the coding apparatus according to Claim 4, the data prediction unit outputs block layout information indicating a layout pattern of various prediction blocks in the coding block. Therefore, it is possible to recognize the arrangement pattern of the prediction error blocks in the coding block at the time of decoding.

According to the encoding apparatus of the present invention (Claim 6), in the encoding apparatus for predictively encoding the image data for each frame using the inter-frame pixel value correlation, the image data corresponding to the frame to be processed The prediction process to predict
A first data predicting unit that generates first predicted image data by performing each first prediction block composed of a plurality of pixels that divide the frame, and image data of the processing target frame, a plurality of first prediction blocks. A first predictive encoding unit having a first data encoding unit that encodes using the first predictive image data corresponding to the first encoded block for each first encoded block consisting of A second data prediction unit that performs a prediction process of predicting image data corresponding to a frame to be processed for each second prediction block including a plurality of pixels that divide the frame, and generates second predicted image data;
Second data that encodes the image data of the processing target frame for each second encoded block that partitions the second prediction block, using the second predicted image data that corresponds to the second encoded block A second predictive coding unit having a coding unit and the image data is supplied to one of the first predictive coding unit and the second predictive coding unit according to a command signal from the outside. A predictive coding process that performs highly accurate motion compensation for image data and a predictive coding process that performs normal accuracy compensation for image data according to the image size to be handled, etc. You can switch between processing and.

According to the encoding method of the present invention (claim 7), there is provided an encoding method for predictively encoding image data frame by frame using pixel value correlation between frames, which corresponds to a frame to be processed. A prediction step of performing prediction processing for predicting image data to be performed for each prediction block composed of a plurality of pixels that partition the frame, and image data of the processing target frame from the plurality of prediction blocks. Each coding block includes a coding step of coding using prediction image data corresponding to the coding block, and thus fine motion detection can be performed and coding efficiency can be improved. It is possible to improve.

According to the present invention (claim 8), in the coding method according to claim 7, the predicting step includes a step of generating predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes. , Of the prediction image data corresponding to a plurality of types of prediction blocks having different shapes,
A step of selecting the prediction image data of the prediction block with the highest coding efficiency for the coding block,
And a step of generating predicted image data corresponding to the coded block based on the predicted image data of the selected prediction block. Therefore, a highly accurate motion matching the motion of the image within the frame is performed. The detection can be performed, and the coding efficiency can be improved.

According to the present invention (claim 9), in the encoding method according to claim 8, the predicting step is performed on any shape of predictive image data corresponding to a plurality of types of prediction blocks having different shapes. Since the method includes a step of generating block identification information indicating whether the predicted image data of the prediction block is selected, the shape of the prediction block can be recognized at the time of decoding.

According to the present invention (claim 10), claim 7
In the encoding method described above, the predicting step is a step of generating predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes, and an arrangement pattern of the various predictive blocks in the encoding block, The step of determining the highest encoding efficiency for the encoded block, and the predicted image data corresponding to the plurality of types of predictive blocks having different shapes, the arrangement pattern of the various predictive blocks in the encoded block is determined. It is characterized in that it includes a step of generating predicted image data corresponding to the above-mentioned coded blocks in combination so as to form an arrangement pattern, so that highly accurate motion detection that matches the shape and motion of a moving image in a frame can be performed. It is possible to improve the coding efficiency.

According to the present invention (claim 11), claim 1
In the encoding method described in 0, the prediction step includes a step of generating block arrangement information indicating an arrangement pattern of various prediction blocks in the encoded block. The arrangement pattern of prediction error blocks can be recognized.

According to the decoding device of the present invention (claim 12), the coded data obtained by predictively coding the image data is predictively decoded for each frame by using the pixel value correlation between the frames. A decoding device, a data prediction unit that performs prediction processing for predicting image data corresponding to a frame to be processed for each prediction block including a plurality of pixels that divide the frame, and generates prediction image data, And a data decoding unit that decodes the coded data of the processing target frame for each decoding block including a plurality of prediction blocks using the prediction image data corresponding to the decoding block. Therefore, it is possible to appropriately decode the encoded data string that has been subjected to the encoding processing by performing the fine motion detection, and from the encoded data encoded with high encoding efficiency, the good image quality can be obtained. Images can play.

According to the present invention (claim 13), claim 1
In the decoding device according to 2, the data prediction unit includes a data generation unit that generates prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, and when performing the predictive coding of the image data, Prediction corresponding to a prediction block of a predetermined shape based on block identification information indicating which prediction image data of which prediction block has been selected among prediction image data corresponding to a plurality of types of prediction blocks of different shapes Image data is selected, and the predicted image data of the selected prediction block is used to generate the predicted image data corresponding to the decoded block. Therefore, highly accurate motion detection matching the motion of the image is performed. It is possible to satisfactorily decode the image data that has been subjected to the encoding process by performing the above, and from the encoded data encoded with high encoding efficiency, It becomes possible to reproduce an image of the image quality.

According to the present invention (Claim 14), Claim 1
In the decoding device according to 2, the data prediction unit includes a data generation unit that generates prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, and the data prediction unit includes a plurality of types of prediction blocks having different shapes. Corresponding predicted image data is combined based on block arrangement information indicating an arrangement pattern of various prediction blocks in the decoding block to generate predicted image data corresponding to the decoding block. Therefore, it is possible to satisfactorily decode coded data that has been subjected to coding processing by performing highly accurate motion detection that matches the shape and motion of a moving image in a frame, and codes coded with high coding efficiency. It is possible to reproduce an image of good quality from the encoded data.

According to the decoding apparatus of the present invention (Claim 15), the coded data obtained by predictively coding the image data is predictively decoded for each frame by using the pixel value correlation between the frames. A decoding device, which performs a prediction process of predicting image data corresponding to a frame to be processed for each first prediction block including a plurality of pixels that partition the frame, and generates first predicted image data 1 data prediction unit and the coded data of the processing target frame for each first decoding block including a plurality of first prediction blocks,
A first predictive decoding unit having a first data decoding unit for decoding using first predictive image data corresponding to the first decoding block, and predicting image data corresponding to a processing target frame The second data prediction unit that performs the prediction process for each second prediction block including a plurality of pixels that partition the frame to generate second predicted image data, and the encoded data of the processing target frame, A second data decoding unit that decodes using the second predicted image data corresponding to the second decoding block for each second decoding block that partitions the second prediction block. A predictive decoding unit and a selection unit that supplies the encoded data to one of the first predictive decoding unit and the second predictive decoding unit according to a command signal from the outside are provided. Since it is a feature, the image Etc. Depending on's can be switched and predictive decoding processing of performing highly accurate motion compensation, and predictive decoding process to compensate for normal accuracy for the image data for the image data.

According to the decoding method of the present invention (claim 16), the coded data obtained by predictively coding the image data is predictively decoded for each frame using the pixel value correlation between frames. A decoding method, a prediction step of performing prediction processing of predicting image data corresponding to a frame to be processed for each prediction block composed of a plurality of pixels that divide the frame, and generating the predicted image data; A decoding step of decoding the coded data of the target frame for each decoding block including a plurality of prediction blocks using the prediction image data corresponding to the decoding block, It is possible to satisfactorily decode a coded data string that has been coded by performing various motion detections, and it is possible to obtain good image quality from coded data that is coded with high coding efficiency. Images can play.

According to the present invention (claim 17), claim 1
6. The decoding method according to 6, wherein the prediction step includes a step of generating prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, and a plurality of types having different shapes at the time of predictive coding of the image data. Of the prediction image data corresponding to the prediction block, the prediction image data corresponding to the prediction block of the predetermined shape is selected based on the block identification information indicating which prediction image data of the prediction block of which shape is selected. It is characterized by including a step and a step of generating predicted image data corresponding to the above-mentioned decoded block by the predicted image data of the selected predicted block. Image data that has undergone high motion detection and encoding processing can be decoded well, and high encoding efficiency is achieved. From encoded data,
It becomes possible to reproduce an image of good quality.

According to the present invention (claim 18), claim 1
6. In the decoding method according to 6, the prediction step includes a step of generating prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, and a prediction image data corresponding to the plurality of types of prediction blocks having different shapes. To
In the frame, since it includes a step of combining based on block arrangement information indicating an arrangement pattern of various prediction blocks in the decoding block to generate prediction image data corresponding to the decoding block. It is possible to satisfactorily decode image data that has been subjected to encoding processing by performing highly accurate motion detection that matches the shape and movement of the moving image of, and from encoded data that is encoded with high encoding efficiency, It becomes possible to reproduce an image of good quality.

[Brief description of drawings]

FIG. 1 is an encoding device 100 according to a first embodiment of the present invention.
It is a block diagram for explaining a.

FIG. 2 is a schematic diagram illustrating a predictive encoding process of the encoding device 100a according to the first embodiment, showing correspondence between a frame corresponding to input image data and a prediction error block.

FIG. 3 is an encoding device 100 according to a second embodiment of the present invention.
It is a block diagram for demonstrating b.

[Fig. 4] Fig. 4 is a schematic diagram illustrating a predictive coding process of the coding device 100b according to the second embodiment, which illustrates an arrangement pattern of 4x4 pixel blocks and 2x8 pixel blocks forming an 8x8 pixel block. Three examples (Fig. (A)-(c)) are shown.

FIG. 5 is an encoding device 100 according to a third embodiment of the present invention.
It is a block diagram for explaining c.

FIG. 6 is an encoding device 100 according to a fourth embodiment of the present invention.
It is a block diagram for explaining d.

FIG. 7 is a recording device 100e according to a fifth embodiment of the present invention.
It is a block diagram for explaining.

FIG. 8 is a recording device 100f according to a sixth embodiment of the present invention.
It is a block diagram for explaining.

FIG. 9 is a decoding device 100 according to a seventh embodiment of the present invention.
It is a block diagram for explaining g.

FIG. 10 is a decoding device 10 according to an eighth embodiment of the present invention.
It is a block diagram for explaining 0h.

FIG. 11 is a decoding device 10 according to a ninth embodiment of the present invention.
It is a block diagram for explaining 0i.

FIG. 12 is a reproduction device 10 according to Embodiment 10 of the present invention.
It is a block diagram for explaining 0j.

FIG. 13 is a reproduction device 10 according to an eleventh embodiment of the present invention.
It is a block diagram for explaining 0k.

FIG. 14 is a diagram for explaining a mobile phone according to a twelfth embodiment of the present invention.

FIG. 15 is a diagram for explaining a data storage medium (FIGS. (A) and (b)) storing a program for executing the apparatus of each of the embodiments by a computer system, and the computer system (FIG. (C)). FIG.

[Fig. 16] Fig. 16 is a schematic diagram for explaining a conventional MPEG encoding system, and shows a frame to which interframe predictive encoding processing is applied.

FIG. 17 is a block diagram illustrating an encoding device 700 using a conventional MPEG encoding method.

[Fig. 18] Fig. 18 is a schematic diagram for explaining a predictive coding process in the conventional MPEG coding system, in which a unit of motion compensation is a 16x16 pixel block and a unit of DCT process is 8.
The correspondence with a × 8 pixel block is shown.

FIG. 19 is a block diagram illustrating a decoding device 800 using a conventional MPEG encoding method.

[Explanation of symbols]

100a to 100d Encoding device 100e, 100f Recording device 100g, 100h Decoding device 100i to 100k Reproducing device 101, 112, 305 4x4 block divider 102, 202, 702 Subtractor 103, 232 8x8 block configurator 104 , 704 DCT device 105, 705 Quantizer 106, 206, 706, 802 Variable length encoder 107, 304, 707, 803 Inverse quantizer 108, 303, 708, 809 Inverse DCT device 109, 209, 306, 405 , 709, 805 Adder 110, 308, 710, 807 Reference frame memories 111, 211, 307, 404, 711, 806 Motion compensation section 123, 723 4 × 4 prediction error block quantization coefficient 124, 224, 330, 724 , 830 Motion vectors 126, 226, 241 331,431,726,8
31 4 × 4 prediction reference block motion compensation data 127, 127a, 127b, 127c, 127d prediction error block data 128, 324, 728, 824 8 × 8 prediction error block decoded data 129, 425 4 × 4 prediction error block Decoded data 130, 340, 440, 730, 840 4 × 4 pixel block decoded data 131, 318 Reference frame data 201 2 × 8 block divider 221 2 × 8 pixel block data 227, 227a, 227b, 227c, 227d 2
× 8 prediction error block data 229 motion compensation block decoded data 231 evaluator 233 decoding block divider 234 selectors 240,410 motion compensation mode identification signals 301,401,801 code classifier 302 variable length decoder 402 prediction block Pattern decoder 403 Block divider 1100 Resolution converters 1101, 1201, 1301, 1401, 1501,
1601, 1701 selectors 1102, 1302, 1402, 1502 small block motion compensation coding unit (first coding unit) 1103, 1303, 1403, 1503 MPEG coding unit (second coding unit) 1104, 1304 1404, 1504 Coded identification code generation section 1200 Coded identification code decoding section 1202, 1602, 1702 Small block motion compensation decoding section (first decoding section) 1203, 1603, 1703 MPEG decoding section (second decoding 1405 recording processing unit 1406, 1606 hard disk 1407, 1607, 1706 DVD 1605, 1705 playback processing unit 1800 mobile phone 1801 antenna 1802 signal processing unit 1803 wireless communication unit 1804 display control unit 1805 audio input / output unit 1806 liquid crystal panel (LCD) ) 18 7 Speaker 1808 Microphone BL4a to BL4d 4 × 4 pixel block BL8 8 × 8 pixel block BLp4a to BLp4d 4 × 4 Prediction error block Bs1 to Bs4, Bsa, Bsb Encoded data string Bs4a, Bs4b, Bs5 Recorded data Cid1 to Cid4 Encoded Identification code Cs Computer system Did2 Encoding identification code F Frame FD Floppy disk FDD Floppy disk drive Id Input image data Im Media designation signal Ir Encoding rate designation signal Is Resolution designation signal RId Reproduced image data

   ─────────────────────────────────────────────────── ─── Continued front page    F-term (reference) 5C059 KK00 MA00 MA05 MA22 MA23                       MC11 MC38 ME01 NN01 NN21                       NN28 PP04 RC11 RF04 SS10                       SS11 TA12 TC25 UA02 UA05                 5J064 AA01 BA04 BA09 BA16 BB03                       BC01 BC08 BC16 BC25 BC27                       BD02 BD03

Claims (18)

[Claims] 1. A coding apparatus for predictively coding image data for each frame using pixel value correlation between frames, wherein a prediction process for predicting image data corresponding to a frame to be processed is divided into A data predicting unit that generates prediction image data by performing each prediction block including a plurality of pixels, and the image data of the processing target frame, for each coding block including a plurality of prediction blocks, corresponding to the coding block. And a data encoding unit that encodes using the predicted image data to be encoded. 2. The encoding device according to claim 1, wherein the data prediction unit has a data generation unit that generates prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, Of the prediction image data corresponding to different types of prediction blocks, the prediction image data of the prediction block having the highest coding efficiency with respect to the coding block is selected, and the prediction image data corresponding to the coding block is generated. An encoding device characterized by being. 3. The encoding device according to claim 2, wherein the data prediction unit selects prediction image data of a prediction block of any shape among prediction image data corresponding to a plurality of types of prediction blocks having different shapes. An encoding device, which outputs block identification information indicating whether or not the process has been performed. 4. The encoding device according to claim 1, wherein the data prediction unit has a data generation unit that generates prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, Prediction image data corresponding to different types of prediction blocks are combined so that the arrangement pattern of the various prediction blocks in the encoding block becomes an arrangement pattern that maximizes the encoding efficiency of the encoding block. An encoding device for generating predicted image data corresponding to. 5. The encoding device according to claim 4, wherein the data prediction unit outputs block arrangement information indicating an arrangement pattern of various prediction blocks in the encoded block. apparatus. 6. An encoding apparatus for predictively encoding image data for each frame using inter-frame pixel value correlation, wherein a predicting process for predicting image data corresponding to a frame to be processed is performed by a plurality of sections for dividing the frame. First composed of pixels
A first data predicting unit that performs first prediction image data by performing each prediction block and image data of the processing target frame for each first coding block including a plurality of first prediction blocks A first predictive encoding unit having a first data encoding unit that encodes using first predictive image data corresponding to one encoded block, and prediction for predicting image data corresponding to a processing target frame The process is performed by a second pixel consisting of a plurality of pixels that divide the frame.
A second data prediction unit that performs second prediction image data by performing each prediction block, and image data of the processing target frame for each second coding block that divides the second prediction block A second predictive coding unit having a second data coding unit for coding using the second predictive image data corresponding to the coding block, and the image data according to a command signal from the outside. An encoding device comprising: a selection unit that supplies one of the first prediction encoding unit and the second prediction encoding unit. 7. A coding method for predictively coding image data for each frame by using pixel value correlation between frames, wherein a predicting process for predicting image data corresponding to a frame to be processed is performed by dividing the frame. A prediction step of generating predicted image data by performing each prediction block composed of a plurality of pixels, and the image data of the processing target frame is associated with each coding block composed of a plurality of prediction blocks. And a coding step of coding using predicted image data. 8. The encoding method according to claim 7, wherein the prediction step includes a step of generating prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, and a plurality of types of prediction having different shapes. Of the prediction image data corresponding to the block, a step of selecting the prediction image data of the prediction block having the highest coding efficiency with respect to the coding block, and the prediction image data of the selected prediction block And a step of generating corresponding predicted image data. 9. The encoding method according to claim 8, wherein in the predicting step, predictive image data of a predictive block of any shape is selected from predictive image data corresponding to a plurality of kinds of predictive blocks having different shapes. An encoding method including a step of generating block identification information indicating that. 10. The encoding method according to claim 7, wherein the prediction step includes a step of generating prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, and the various predictions in the coding block. A step of determining a block arrangement pattern so that the coding efficiency with respect to the coding block is the highest; and predictive image data corresponding to a plurality of types of prediction blocks having different shapes, the various prediction blocks in the coding block. And generating the predicted image data corresponding to the coded block by combining the arrangement patterns of the above-described arrangement patterns so that the arrangement pattern becomes the determined arrangement pattern. 11. The encoding method according to claim 10, wherein the predicting step includes a step of generating block arrangement information indicating an arrangement pattern of various prediction blocks in the encoding block. Method. 12. A decoding device for predictively decoding coded data obtained by predictively coding image data for each frame using pixel value correlation between frames, the image corresponding to a frame to be processed. A data prediction unit that performs prediction processing for predicting data for each prediction block composed of a plurality of pixels that divide the frame to generate predicted image data, and encoded data of the processing target frame from the plurality of prediction blocks. And a data decoding unit that performs decoding using the predicted image data corresponding to each of the decoding blocks. 13. The decoding device according to claim 12, wherein the data prediction unit includes a data generation unit that generates prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes. In predictive coding of, a predetermined shape is calculated based on block identification information indicating which shape of the predictive image data is selected from the predictive image data corresponding to a plurality of types of predictive blocks having different shapes. Decoding device for selecting prediction image data corresponding to the prediction block, and generating prediction image data corresponding to the decoding block based on the prediction image data of the selected prediction block. 14. The decoding device according to claim 12, wherein the data prediction unit has a data generation unit that generates prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, Prediction image data corresponding to different types of prediction blocks are combined based on block arrangement information indicating an arrangement pattern of various prediction blocks in the decoding block to generate prediction image data corresponding to the decoding block And a decoding device. 15. A decoding device for predictively decoding coded data obtained by predictively coding image data for each frame using pixel value correlation between frames, the image corresponding to a frame to be processed. A prediction process for predicting data is performed by a first process including a plurality of pixels that divide the frame.
A first data predicting unit that performs first prediction image data by performing each prediction block and the encoded data of the processing target frame for each first decoding block including a plurality of first prediction blocks The first corresponding to the first decoding block
A first predictive decoding unit having a first data decoding unit for decoding using the predicted image data of, and a prediction process for predicting image data corresponding to a processing target frame, 2nd consisting of pixels
A second data predicting unit that performs second prediction image data by performing each prediction block and coded data of the processing target frame for each second decoding block that divides the second prediction block The second corresponding to the second decoding block
Second predictive decoding unit having a second data decoding unit for decoding using the predicted image data of No. 1, and the encoded data is converted into the first predictive decoding unit according to a command signal from the outside. A decoding device comprising: a conversion unit and a selection unit that supplies the second predictive decoding unit. 16. A decoding method for predictively decoding coded data obtained by predictively coding image data for each frame using pixel value correlation between frames, the image corresponding to a frame to be processed. A prediction step of generating prediction image data by performing a prediction process of predicting data for each prediction block composed of a plurality of pixels that partition the frame, and the encoded data of the processing target frame is composed of a plurality of prediction blocks. And a decoding step of decoding, for each decoding block, using predicted image data corresponding to the decoding block. 17. The decoding method according to claim 16, wherein the predicting step includes a step of generating predictive image data corresponding to each of a plurality of types of predictive blocks having different shapes, and predictive encoding of the image data. At this time, among the prediction image data corresponding to a plurality of types of prediction blocks of different shapes, the prediction block of a predetermined shape is supported based on the block identification information indicating which prediction block of the prediction image data of the shape is selected. A decoding method, comprising: selecting the predicted image data to be processed; and generating the predicted image data corresponding to the decoding block from the predicted image data of the selected prediction block. 18. The decoding method according to claim 16, wherein the prediction step includes a step of generating prediction image data corresponding to each of a plurality of types of prediction blocks having different shapes, and a plurality of types of prediction having different shapes. Generating predicted image data corresponding to the decoded block by combining predicted image data corresponding to the block based on block arrangement information indicating an arrangement pattern of various predicted blocks in the decoded block. Decoding method characterized by.