JPH07123271A - Method for compressing and expansting picture data - Google Patents

Abstract

PURPOSE:To prevent a block distortion on a screen from being sensed by means of the integration effect of human eyes. CONSTITUTION:Picture data by one frame from a camera 1 stored in memory 3 in a block processing circuit 2 are block-processed while the position to be block-processed is revised under the control of a control circuit 4. Then the resulting data are fed to a transmission system or recorded on a recording medium 8 via a DCT circuit 5, a quantization circuit 6 and a VLC circuit 7 and then the data are converted again into block-processed picture data via an IVLC circuit 9, an inverse quantization circuit 10 and an IDCT circuit 11 and stored in memory 13 in an inverse block processing circuit 12. The picture data in the memory 13 are decoded into picture data by one frame under the control of a control circuit 14 and the data are outputted to a monitor 15, in which the data are displayed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a compression method and an expansion method for image data taken by an image pickup device.

[0002]

2. Description of the Related Art In recent years, a method of compressing image data and outputting it to an image display device via a recording medium (so-called storage medium) or a transmission system has been proposed. in this way,
When compressing image data, block coding, in particular, discrete cosine transform coding (DCT) is used. This discrete cosine transform coding divides the image data into small blocks by utilizing the fact that the image data is smooth and has almost no high frequency components, and the image data is divided into small blocks on the frequency axis. By converting to data,
The amount of data recorded on the recording medium or output to the transmission system can be compressed.

FIG. 6 shows an image data string of one frame. Normally, one frame of image data is transmitted in the order of arrows in FIG. 6, that is, in the raster scan order. When rearranging the image data for one frame input in the raster scan order and converting it into, for example, a block of 8 rows and 8 columns (8 × 8), as shown in FIG. Is output, and then the 8 × 8 block B is output. In this way, the image data which has been sequentially divided into blocks of 8 × 8 blocks is output, and finally 8 × 8 blocks are output.
By outputting the block Z of, the image data for one frame is divided into blocks.

Blocked image data is transmitted as a compressed image data to a transmission system or recorded on a recording medium via a DCT circuit, a quantization circuit, and a VLC (Variable Length Coding) circuit. .

Compressed image data transmitted to a transmission system or recorded on a recording medium is passed through an IVLC (variable length decoding) circuit, an inverse quantization circuit, and an IDCT (inverse discrete cosine transform encoding) circuit. It is extended. At this time, the block image data is output from the IDCT circuit in the order of block A, block B, ..., Block Z. The blocked image data is input to the deblocking circuit, deblocked, and output in the raster scan order before being compressed.

The compression and decompression of image data as described above is performed for all frames of an image.

Next, the compression and decompression of image data will be specifically described. For example, there is one-dimensional image data as shown in FIG. 8, and the one-dimensional image data is divided into blocks of eight image data and converted into block A, block B, and block C, respectively. To do. After this, the DCT is applied to these blocks A, B, and C, respectively.
Is applied and converted into data on the frequency axis, that is, a DC component and an AC component.

Further, the data on the frequency axis is quantized. This quantization is to transform the quantization step of the data on the frequency axis. Specifically, for example, when the predetermined data is represented by a 10-bit integer, the predetermined data is divided by 4 and the decimal point and below are rounded off. As a result, this predetermined data is represented by an 8-bit integer. That is, by performing quantization, 10-bit data can be compressed into 8 bits. The quantized data is further subjected to variable length coding (VLC) and coded.

When decompressing the compressed image data, variable length decoding (IV
LC) is applied, followed by dequantization. Specifically, for example, the above 8-bit data is multiplied by 4 to be 10-bit data. The dequantized data is the data on the frequency axis of each block, that is, the DC component and the AC component.

Thereafter, the dequantized data on the frequency axis is subjected to IDCT and converted into block image data. That is, the image data before being compressed is restored.

In the above inverse quantization, the quantized data is simply quadrupled by simply shifting the 8-bit data by 2 bits and adding 0 to the lower 2 bits. is there. Therefore, between the 10-bit data before being quantized and the 10-bit data after being dequantized, an error (shift) occurs in the lower 2 bits. That is, since the 10-bit data is forcibly reduced to 8 bits by the quantization, even if the inverse quantization is performed thereafter, the original data before the quantization cannot be accurately restored.

Here, for example, when the blocked image data of FIG. 8 is compressed and expanded as described above,
Only one DC component in the data on the frequency axis corresponding to the block B of FIG. 8 has an error of 1 due to the quantization and the inverse quantization, and the error in the data on the frequency axis corresponding to the blocks A and C in FIG. It is assumed that the DC component has no error. After this, IDCT is applied to the data on the frequency axis of this block B.
As shown in FIG. 9, the image data obtained by performing the process becomes larger than the original image data before the quantization by one.
However, IDCT is added to the data on the frequency axis of blocks A and C.
The image data obtained by applying is restored to the original image data.

[0013]

[Problems to be Solved by the Invention]

By the way, when the image data is divided into blocks and compressed and then the compressed data is expanded, a step is formed at the boundary between blocks in each frame, for example, at the boundary between block A and block B in FIG. The human eye perceives this step as block distortion.

When the image data compression method and decompression method as described above are used, a constant block is always formed in all frames. That is, for all the frames, for example, the image data for one frame is converted into blocks A to Z as shown in FIG. The block distortion will always be sensed in a certain part (boundary between blocks) of the.

Further, the image data is divided into macroblocks, the motion vector between the frames of each macroblock is obtained, and then the difference with another frame moved by this motion vector is calculated. DC
An image data compression method for performing T is also known. This is a compression method of image data called a hybrid method using motion compensation and DCT. In this image data compression method, in general,
The size of the macroblock for which the motion vector is calculated, and DC
It is different from the size of the block that performs T. Since motion compensation does not operate effectively for this macroblock as well, block distortion may occur.

In view of the above situation, the present invention provides a method of compressing image data and a method of decompressing image data in which block distortion is not perceived by human eyes.

[0018]

A method of compressing image data according to the present invention solves the above-mentioned problem by changing the block cut-out position of the image data for each frame when the image data is divided into blocks.

Further, in the image data decompression method according to the present invention, when the image data is divided into blocks, the block cut-out position of the image data is changed for each frame to form a block, and the compressed image data is expanded. After that, when performing inverse blocking on the image data of each block, the above problem is solved by controlling the extraction of the image data of each frame according to the change in the block cutout position of the image data.

[0020]

In the present invention, when compressing image data,
When the image data is divided into blocks, the block cut-out position of the image data is shifted for each frame and the divided image data is compressed, and when the compressed data is expanded, the compressed image data is expanded. After that, by controlling the image data for each block so as to generate image data for one frame, it is possible to shift the position where the block distortion of the image data occurs.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a schematic configuration of an apparatus using the image data compression method and decompression method according to the present invention.

Image data taken by the camera 1 shown in FIG. 1 is input to the blocking circuit 2 in the same raster scan order as in the conventional case and stored in the memory 3 in the blocking circuit 2. Under the control of the control circuit 4, the image data stored in the memory 3 is changed into blocks by changing the position where the image data is blocked for each frame.

Specifically, 1 stored in the memory 3 is stored.
When the image data for a frame is divided into blocks under the control of the control circuit 4 and the image data for an odd number of frames is divided into blocks, for example, as shown in FIG.
From the upper left pixel, as a block of 8 rows and 8 columns (8 × 8), block A1, block B1, ..., Block Z
Blocking is performed in order of 1. Therefore, the image data for the four pixels from the right side and the four pixels from the lower side, which are shown by the diagonal lines in FIG. Further, when the image data of the even frames is divided into blocks, for example, the blocks A2, B2, ...
Blocking is performed in the order of the block Z2. Therefore, the image data of 4 pixels from the upper side and 4 pixels from the left side of the even frame are not divided into blocks. In this way, in the odd-numbered frame and the even-numbered frame, the position of the pixel on the screen to be blocked is changed to perform blocking. Here, the area indicated by the thick dotted line in FIGS. 2 and 3 is
The effective display screen area mentioned later is shown.

These block image data are D
The data is output to a CT (discrete cosine transform coding) circuit 5, converted into data on the frequency axis, and then supplied to a quantization circuit 6. The quantization circuit 6 changes the quantization step and quantizes the data on the frequency axis. The data quantized by the quantization circuit 6 is sent to a VLC circuit (variable length coding) circuit 7 and coded, and then transmitted to a transmission system or a recording medium by a transmission system or a recording medium 8. Will be recorded.

The data transmitted to the transmission system or recorded on the recording medium is sent to the IVLC (Variable Length Decoding) circuit 9, decoded, and sent to the dequantization circuit 10. The inverse quantization circuit 10 inversely quantizes the transmitted quantized data, and the inversely quantized data is supplied to an IDCT (inverse discrete cosine transform encoding) circuit 11. This I
In the DCT circuit 11, the IDCT is applied to the dequantized data.
And convert it to data on the frequency axis.

Data on the frequency axis output from the IDCT circuit 11 is stored in the memory 1 in the deblocking circuit 12.
3 is sequentially input for each block. Specifically, the block image data of the odd-numbered frame is
1, block B1, ..., Block Z1 in order, and even-numbered framed image data are input to the memory 3 in the order of block A2, block B2 ,. Remembered.

Thereafter, under the control of the control circuit 14 in the deblocking circuit 12, when the image data for each block stored in the memory 13 is read out, dummy data is inserted and rasterized as image data for one frame. The images are rearranged in the order of scan, and are output and displayed on the monitor 15 which is a display device.

In this case, the image data sent to the decompression side via the transmission system or the recording medium 8 is more likely than the image data input from the camera 1 for the odd frames as shown in FIG. 4 pixels from the left side and 4 pixels from the lower side of
As shown in FIG. 3, there are four pixels from the right side of the screen and four pixels from the upper side. Therefore, the image effectively displayed on the monitor 15 is an overlapping portion of the transmitted or recorded image data in the image data of each of these frames, and is smaller than the image captured by the camera 1. , And the image in the image frame indicated by the thick dotted line in FIGS. 2 and 3. As described above, by reducing the image frame (effective display screen area) displayed on the monitor 15, both odd and even frames can be reproduced and displayed.

The operation of the blocking circuit 2 of FIG. 1 is shown in the flowchart of FIG. 4, and the operation of the deblocking circuit 12 of FIG. 1 is shown in the flowchart of FIG.

In the blocking circuit 2, first, in step S1 of FIG. 4, image data is input in the raster scan order and written in the memory 3. Next, step S
In step 2, it is determined whether the image data stored in the memory 3 is an odd frame. If the stored image data is an odd number of frames, the process proceeds to step S3, and as shown in FIG.
The data is divided into 8 blocks, read out, and output in block units. If the image data stored in the memory 3 is an even frame, the process proceeds to step S4 and, as shown in FIG. 3, an image at a position shifted by 4 pixels from the left side of the screen and 4 pixels from the upper side. The data is read by dividing it into 8 × 8 blocks and output in block units.

In the deblocking circuit 12, first, in step S5 of FIG. 5, image data is input in 8 × 8 block units and written in the memory 13.
Next, in step S6, it is determined whether the image data stored in the memory 13 is an odd frame.
If the stored image data is an odd number of frames, the process proceeds to step S7, and when reading the image data from the memory 13, four pixels from the right side of the screen and four pixels from the lower side, which are indicated by the diagonal lines in FIG. After the dummy data is inserted to generate one frame, it is output in the raster scan order in step S9. If the image data stored in the memory 13 is an even frame, the process proceeds to step S8 to read the image data from the memory 13,
Dummy data is inserted into four pixels from the left side and four pixels from the upper side of the screen, which are indicated by the diagonal lines in FIG. 3, to generate one frame, which is output in the raster scan order in step S9.

As described above, in the apparatus using the image data compression method and the decompression method according to the present invention, the odd-numbered frame image data block positions and the even-numbered frame image data block positions are set. change.

When the image data for one frame read from the memory 13 is to be the image data within the image frame (effective display screen area) shown by the thick dotted lines in FIGS. 2 and 3, the inverse blocking circuit is used. 12, the pixel located in the effective display screen area in the odd frame and the even frame is detected from the memory 13 without adding the dummy data as described above, and the pixel is located in the upper left of the effective display screen area. It is also conceivable that the pixels are read in raster scan order and output.

When the image data is displayed on the monitor 15, distortion occurs at the boundary between blocks in each frame. Specifically, in odd-numbered frames, block distortion occurs at the positions indicated by the thin dotted lines in FIG. 2, and in even frames, block distortion occurs at the positions indicated by the thin dotted lines in FIG. However, since the position of the block distortion in the odd frame and the position of the block distortion in the even frame are different, the block effect generated in the odd frame is fooled by the even frame due to the integration effect of human eyes.
The block distortion caused in the even frames is fooled by the odd frames. Therefore, the human eye does not perceive block distortion on the screen.

Even when the image data compression method using macroblocks is used, the positions of the macroblocks in the odd frames and the even frames are different, so that the human eye is deceived and the block distortion can be sensed. Absent.

In the above-described embodiment, the odd-numbered frame image data is blocked as shown in FIG. 2 and the even-numbered frame image data is blocked as shown in FIG. Alternatively, the image data may be blocked by using one of the image data shown in FIGS. 2 and 3 by using a random number. By using this method, the block formation of the image data shown in FIG. 2 and the block formation of the image data shown in FIG. 3 occur at random, so that it is possible to further deceive the human eye. However, at this time, information indicating which of the image data blocks shown in FIGS. 2 and 3 has been selected for each frame is also transmitted to the transmission system or recorded in the recording medium, and when decompressing the compressed data. Appropriate deblocking is performed based on this information.

Further, in the above-mentioned embodiment, the two types of image data blocks shown in FIGS. 2 and 3 are used, but the image data blocks may be divided into three or more types.

Furthermore, the image data compression method and decompression method of the present invention can also be applied to a system that handles stereoscopic images using right-eye image data and left-eye image data. That is, for the image data for the right eye, the image data shown in FIG. 2 is blocked to compress the image data, and for the image data for the left eye, the image data shown in FIG. 3 is blocked. Go and compress the image data. Thereafter, when the compressed right-eye image data and left-eye image data are expanded and displayed on a stereoscopic image display device, the block distortion position of the image data viewed by the right eye and the image viewed by the left eye. Since the location of the block distortion in the data is different, the human eye is fooled into not seeing the block distortion on the screen.

[0039]

As is clear from the above description, the image data compression method according to the present invention changes the block cut-out position of the image data for each frame when the image data is divided into blocks. Due to the integration effect of the human eye, block distortion on the screen can be made insensitive.

Further, in the image data decompression method according to the present invention, when the image data is divided into blocks, the block cut-out position of the image data is changed for each frame to form a block, and the compressed image data is expanded. After that, when performing inverse blocking on the image data of each block, the integration effect of human eyes is controlled by controlling the extraction of the image data of each frame according to the change of the block cutout position of the image data. This makes it possible to prevent the block distortion on the screen from being sensed.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of an apparatus using a compression method and an expansion method for image data according to the present invention.

FIG. 2 is a diagram showing how image data of odd-numbered frames is divided into blocks.

FIG. 3 is a diagram showing how image data of even frames is divided into blocks.

FIG. 4 is a flow chart showing the operation of the blocking circuit of FIG.

5 is a flow chart showing the operation of the deblocking circuit of FIG. 1. FIG.

FIG. 6 is a diagram showing a raster scan order of image data on an image captured by a camera.

FIG. 7 is a diagram showing how the image data of FIG. 6 is divided into blocks.

FIG. 8 is a block A of block image data,
It is a figure which shows B and C with one-dimensional image data.

9 is a diagram showing a state after blocks A, B, and C shown in FIG. 8 have been compressed and expanded.

[Explanation of symbols]

 1 ... Camera 2 ... Blocking circuit 3, 13 ... Memory 4, 14 ... Control circuit 5 ... DCT circuit 6 ... Quantization circuit 7 ... ..VLC circuit 8 ... Transmission or recording medium 9 ... IVLC circuit 10 ... Dequantization circuit 11 ... IDCT circuit 12 ... Deblocking circuit 15 ... ··monitor

Claims (2)

[Claims] 1. A method of compressing image data in which image data is divided into blocks and compression processing is performed for each block, wherein when the image data is divided into blocks, a block cutout position of the image data is changed for each frame. Characteristic image data compression method. 2. An image data decompression method for dividing image data into blocks, decompressing the image data compressed for each block, and deblocking the image data for each block. At the time of blocking, the block cutout position of the image data is changed for each frame to be blocked, and after decompressing the compressed image data, when performing deblocking on the image data of each block, A method for decompressing image data, characterized in that extraction of image data for each frame is controlled according to a change amount of a block cut-out position of the image data.