JPH07240943A - Stereoscopic image encoding method - Google Patents

Abstract

PURPOSE:To reproduce an image on one side with satisfactory quality at least even at the time of the special reproduction of a VTR such as search by encoding a main image and a sub image so as to fix the amount of codes for each combination of small fixed length unit block pairs corresponding to the positions of the main image and the sub image. CONSTITUTION:A main image Lch and a sub image Rch inputted from a camera or the like are respectively stored in a frame memory 2. Next, a left/right parallax vector and a motion vector are calculated by a parallax compensation/ motion compensation circuit 15 and the parallax/motion of the main image is compensated corresponding to the sub image. Afterwards, the difference between the source image of the sub image and the parallax/motion compensated main image is calculated by a subtracter 16. Then, the amount of codes corresponding to the source image of the main image and the predictive residual of the sub image calculated by the subtracter 16 is allocated. Concerning both the images, the amount of codes is fixed for each combination of small fixed length unit blocks made correspondent in the manner of position.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a stereo image coding method, and more particularly to a stereo image coding method suitable for use in a digital VTR.

[0002]

2. Description of the Related Art Conventional stereo image coding methods include
There are a method of performing coding using the left-right correlation and a method of performing independent coding on the left and right without using the left-right correlation. In the conventional method of encoding using the correlation between the left and right, without encoding to a fixed bit amount in small fixed length units, from the accumulation condition of the buffer memory, the characteristics of each block, etc.,
The quantization parameter is obtained and encoded. Among the conventional methods of independently encoding left and right without utilizing left-right correlation, there is one in which left and right images are independently encoded in small fixed length units.

[0003]

In the conventional method of performing coding using the left-right correlation, the coding is not completed in small fixed length units because the small fixed length units are not coded in a fixed bit amount. Therefore, when data can be obtained only intermittently over tracks, such as a special reproduction of a digital VTR such as a search, there is a problem that a reproduced image cannot be obtained on either side. Further, in the conventional method in which the left and right images are independently encoded in small fixed length units, special reproduction is possible, but there is a problem that the encoding efficiency is poor because the correlation between the left and right images is not used.

According to the present invention, at the time of special reproduction of a VTR such as a search, at least one side image can be reproduced with good quality, and the encoding efficiency is higher than that of encoding independently in small fixed length units. It is an object of the present invention to provide a stereo image coding method which is high and which can obtain a better image during normal reproduction.

[0005]

A first stereo image coding apparatus according to the present invention divides a main image and a sub-image into small fixed-length unit blocks, and small fixed-length unit blocks corresponding to the positions of both images. The main image and the sub-image are coded so that the code amount for each combination of pairs becomes constant.

A second stereo image coding apparatus according to the present invention performs intra-frame coding on a main image, performs parallax compensation and motion compensation on the main image, and performs sub-parallax compensation and motion compensation on the sub-image. In a stereo image coding method in which the difference from the main image is coded as prediction residual data of the sub image, the main image and the sub image are divided into small fixed length unit blocks, and the small fixed length corresponding to the position of both images is fixed. Prediction residual data of the main image and the sub-image is encoded so that the code amount for each combination of unit block pairs is constant, and is constant for each combination of small fixed length unit block pairs corresponding to the position of both images. It is characterized by outputting the sign of the quantity.

A third stereo image coding apparatus according to the present invention performs intra-frame coding on a main image, performs parallax compensation and motion compensation on a sub image with respect to the main image, and performs parallax compensation and motion compensation on the main image. In a stereo image coding method in which the difference from a sub-image is coded as prediction residual data of the sub-image, the main image and the sub-image are divided into small fixed-length unit blocks, and the small fixed length corresponding to the position of both images is fixed. Prediction residual data of the main image and the sub-image is encoded so that the code amount for each combination of unit block pairs is constant, and is constant for each combination of small fixed length unit block pairs corresponding to the position of both images. It is characterized by outputting the sign of the quantity.

In the second or third stereo image coding apparatus, the code amount distribution of the prediction residual data of the main image and the sub image is, for example, the data amount of the original image of the main image and the prediction residual of the sub image. Based on the difference data amount, the code amount is determined to be constant in the small fixed length unit block for each combination of the small fixed length unit block pairs corresponding to the positions of both images.

[0009]

In the first stereo image coding apparatus according to the present invention, the main image and the sub image are divided into small fixed length unit blocks. Then, the main image and the sub-image are encoded so that the code amount for each combination of the small fixed length unit block pairs corresponding to the position of both images becomes constant.

In the second stereo image coding apparatus according to the present invention, the main image is intra-frame coded. Also, the main image is parallax-compensated and motion-compensated with respect to the sub-image, and the difference between the sub-image and the main image after parallax-compensated and motion-compensated is encoded as prediction residual data of the sub-image. The main image and the sub-image are divided into small fixed length unit blocks. The prediction residual data of the main image and the sub-image are encoded so that the code amount for each combination of the small fixed length unit block pairs corresponding to the position of both images is constant. Then, a certain amount of code is output for each combination of the small fixed length unit block pairs corresponding to the position of both images.

In the third stereo image coding apparatus according to the present invention, the main image is intra-frame coded. In addition, the sub-image is parallax-compensated and motion-compensated with respect to the main image, and the difference between the main image and the sub-image after parallax-compensated and motion-compensated is encoded as prediction residual data of the sub-image. The main image and the sub-image are divided into small fixed length unit blocks. The prediction residual data of the main image and the sub-image are encoded so that the code amount for each combination of the small fixed length unit block pairs corresponding to the position of both images is constant. Then, a certain amount of code is output for each combination of the small fixed length unit block pairs corresponding to the position of both images.

[0012]

Embodiments of the present invention will be described below with reference to FIGS. FIG. 1 shows an encoder. In this example, the left image (Lch) is the main image and the right image (Rch) is the sub image. First, a main image (Lch) and a sub-image (Rc) input from a camera or the like.
h) is stored in the frame memory 1 and the frame memory 2, respectively. Next, in parallax compensation / motion compensation 15,
Left and right parallax vectors and motion vectors are obtained, and the main image is parallax-compensated and motion-compensated with respect to the sub-image.
After that, the subtracter 16 calculates the difference between the original image of the sub-image and the main image after parallax compensation and motion compensation (prediction residual of the sub-image). Then, the code amount distribution calculation circuit 17
Then, the code amount is assigned to the prediction residual of the original image of the main image and the sub-image obtained by the subtracter 16.

The main image and the sub-image are divided into small fixed length unit blocks composed of a plurality of blocks, which are DCT transform units described later. In the code amount distribution calculation circuit 17,
Prediction residuals of the original image and the sub-image of the main image for each small fixed length unit block pair so that the code amount for each combination of the small fixed length unit block pairs corresponding to the position of the main image and the sub image is constant The code amount distribution of data is determined.

The main image is encoded in parallel with the above code amount distribution calculation. At this time, the switch 3 and the switch 7 are switched to the contact a side. The original image of the sub-image is sent to the DCT circuit 4 in a predetermined block unit of, for example, 8 lines × 8 pixels, and is DCT-converted in a block unit. The DCT-transformed data is quantized by the quantization circuit 5. At this time, the quantization parameter used in the quantization circuit 5 is set so that the code amount of the main image fits within the code amount assigned to the original image of the main image by the code amount distribution calculation circuit 17. Controlled by.

The quantized data is variable-length coded together with the quantization parameter in the variable-length coding circuit 6 and stored in the buffer memory 8 via the switch 7. Further, the quantized data is locally decoded by the inverse quantization circuit 10 and the inverse DCT circuit 11 and stored in the frame memory 12. The above processing is continued until one frame of the main image is completed.

When the processing for one frame of the main image is completed, the switches 3 and 7 are switched to the contact b side. In the parallax compensation / motion compensation circuit 13, the decoded image of the main image (hereinafter referred to as the reference image) accumulated in the frame memory 12 is subjected to parallax compensation and motion compensation for the sub image. Then, the difference between the reference image after the parallax compensation and the motion compensation and the original image of the sub image (prediction residual of the sub image) is obtained by the subtractor 14.

The prediction residual data of the sub-picture is DCT-transformed by the DCT circuit 4 and quantized by the quantizing circuit 5 in the same manner as the encoding of the main picture described above. At this time, the quantization parameter used in the quantization circuit 5 is that the code amount of the prediction residual data of the small fixed length block of the sub-image is calculated from the constant code amount determined for the small fixed length unit block pair. It is controlled by the code amount control circuit 18 so as to be a value obtained by subtracting the code amount for the small fixed length unit block of the main image corresponding to the position.

The quantized data is variable length coded by the variable length coding circuit 6 together with the quantization parameter, the disparity vector and the motion vector. The variable-length encoded data is sent to the small fixed length unit configuration circuit 9 together with the main image data stored in the buffer memory 8, and is combined by the small fixed length unit configuration circuit 9 as small fixed length data. Is output.

The above processing is performed for each small fixed length unit block. For example, as shown in FIG. 3, when one frame on the left and right is recorded on 10 tracks of the VTR, and there are two pairs of small fixed-length unit blocks on each track, for example, I0 and P0, I1 in FIG. And P1, I2 and P2, ... Are small fixed length unit block pairs. FIG. 2 shows a decoder.

Similarly to the encoder, the decoder also decodes the sub-image after decoding the main image.
First, the input small fixed length unit code is Intra /
The inter data separation circuit 21 separates the code for the main image and the code for the prediction residual data of the sub image. The code for the prediction residual data of the sub image is accumulated in the buffer memory 26 until the decoding of the main image is completed.

The code for the main image is a variable length decoding circuit 2
2. Decoded by the inverse quantization circuit 23 and the inverse DCT circuit 24. At this time, the quantization parameter decoded by the variable length decoding circuit 22 is sent to the inverse quantization circuit 23 and used for the inverse quantization of the inverse quantization circuit 23. The image data of the decoded main image is temporarily stored in the frame memory 25 and then output.

When the decoding of the code for the main image is completed, the code for the prediction residual data of the sub image is decoded from the buffer memory 26 by the variable length decoding circuit 27, the dequantization circuit 28 and the inverse DCT circuit 29. At this time, the quantization parameter decoded by the variable length decoding circuit 27 is sent to the inverse quantization circuit 28 and used for the inverse quantization of the inverse quantization circuit 28. In addition, the parallax vector and the motion vector decoded by the variable length decoding circuit 27 are used for parallax compensation /
It is sent to the motion compensation circuit 31.

On the other hand, the parallax compensation / motion compensation circuit 31 uses the parallax vector and the motion vector sent from the variable length decoding circuit 27 to transform the image data of the main image stored in the frame memory 25. And motion compensated. Prediction residual data of the decoded sub-image,
The data of the main image after the parallax compensation and the motion compensation are added by the addition circuit 30, the sub image is decoded, and is output via the frame memory 32.

In the encoder of the above embodiment, the difference between the image after the parallax compensation and the motion compensation of the decoded image of the main image with respect to the original image of the sub image and the original image of the sub image is taken, Although the prediction residual of the sub-image has been obtained, by taking the difference between the image after the parallax compensation and motion compensation of the original image of the sub-image with respect to the decoded image of the main image, and the decoded image of the main image, You may make it obtain the prediction residual of a sub-image.

In this case, in order to decode the sub-picture data on the encoder side, the data of the main picture obtained by the decoding is added to the decoded data of the prediction residual data of the sub-picture, and after this addition Parallax compensation and motion compensation may be performed on the main image data obtained by decoding the data of 1. In the above embodiment, the left image is the main image and the right image is the sub image. However, the left image may be the sub image and the right image may be the main image.

[0026]

According to the present invention, the main image and the sub-image are encoded so that the code amount for each combination of the small fixed length unit block pairs corresponding to the position of the main image and the sub-image is constant. Therefore, at the time of special reproduction of the VTR such as search, at least one image can be reproduced with good quality. In addition, since the coding efficiency is higher than the coding in the small fixed length units independently on the left and right, a better image can be obtained during normal reproduction.

[Brief description of drawings]

FIG. 1 is an electric block diagram showing a configuration of an encoder.

FIG. 2 is an electrical block diagram showing a configuration of a decoder.

FIG. 3 is a schematic diagram showing an example of a recording position on a VTR track.

FIG. 4 is a schematic diagram showing a configuration example of a small fixed length unit.

[Explanation of symbols]

 1, 2 and 12 Frame memory 4 DCT circuit 5 Quantization circuit 6 Variable length coding circuit 8 Buffer memory 9 Small fixed length unit configuration circuit 10 Inverse quantization circuit 11 Inverse DCT circuit 13 Parallax compensation / motion compensation circuit 14 Subtractor 15 Parallax compensation / motion compensation circuit 16 Subtractor 17 Code amount distribution calculation circuit 18 Code amount control circuit

Claims (4)

[Claims] 1. A main image and a sub-image are divided into small fixed-length unit blocks, and the main image and the sub-image are arranged so that the code amount is constant for each combination of the small fixed-length unit block pairs corresponding to the position of both images. A stereo image encoding method for encoding a sub-image. 2. A main image is intra-frame coded, the main image is subjected to parallax compensation and motion compensation with respect to the sub image, and the difference between the sub image and the parallax-compensated and motion-compensated main image is the prediction residual of the sub image. In the stereo image coding method of coding as difference data, the main image and the sub-image are divided into small fixed-length unit blocks,
The prediction residual data of the main image and the sub-image are coded so that the code amount for each combination of the small fixed-length unit block pairs corresponding to the positions of both images is constant, and the small residuals corresponding to the positions of both images are small. A stereo image encoding method, which outputs a fixed amount of codes for each combination of fixed-length unit block pairs. 3. A main image is intra-frame coded, the sub image is subjected to parallax compensation and motion compensation with respect to the main image, and the difference between the main image and the parallax compensated and motion compensated sub image is the prediction residual of the sub image. In the stereo image coding method of coding as difference data, the main image and the sub-image are divided into small fixed-length unit blocks,
The prediction residual data of the main image and the sub-image are coded so that the code amount for each combination of the small fixed-length unit block pairs corresponding to the positions of both images is constant, and the small residuals corresponding to the positions of both images are small. A stereo image encoding method, which outputs a fixed amount of codes for each combination of fixed-length unit block pairs. 4. The code amount for each combination of small fixed length unit block pairs corresponding to the position of both images is constant based on the data amount of the original image of the main image and the prediction residual data amount of the sub image. 4. The image coding method according to claim 2, wherein the code amount distribution of the prediction residual data of the main image and the sub image is determined so that.