JPH0646399A - Picture data compression method - Google Patents

Abstract

PURPOSE:To provide a picture data compression method for improving a compression rate at the time of generating linear arrangement data for each picture element block and improving the beauty of reproduced pictures. CONSTITUTION:The compression object pictures of one frame stored in a bit map memory 1 are divided into the blocks of 8X8 picture elements, a first-order differential operator 3 is operated to the structural pixel data of the respective blocks and a sampling direction is decided in the direction where a gradient is small. Then, an optional compression processing is executed to the linear arrangement data obtained by sampling.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data compression method for dividing an image to be compressed in one frame or one field into a plurality of blocks and compressing (quantizing) the image data for each block. , Video conferencing / videophone, devices that have compression / expansion functions to improve data transmission and storage efficiency for image data such as color still images, especially color moving images, and devices that contribute to AVCC (audio / visual computer / communication) integration The present invention relates to a method of compressing image data used in.

[0002]

2. Description of the Related Art The conventional image data compression method is 1
The frame is divided into blocks of 8 × 8 pixels or 16 × 16 pixels, and compression processing is performed on the one-dimensional array data obtained by sampling all blocks in a fixed direction. In a relatively large number of cases, the correlation of each color in a block of pixel data appears remarkably in the vertical direction or the horizontal direction, so that the sampling is performed in a diagonal fixed direction, which is neither of them. This is because, for example, when compressing data obtained by sampling an image having a strong correlation in the vertical direction in the horizontal direction, not only the compression rate decreases but also the loss of data becomes severe depending on the compression method, and the reproduced image This is because an inconvenient striped pattern may appear.

[0003]

However, in the above-mentioned conventional technique, the compression processing is performed on the one-dimensional array data obtained by uniformly sampling all blocks in a fixed direction. , Some blocks may have a correlation in a direction different from the sampling direction. In such a case, in particular, when there is a strong correlation in the direction orthogonal to the sampling direction, not only the compression rate in the block decreases but also the loss of data becomes severe and the reproduced image becomes unsightly. An object of the present invention is to eliminate the above-mentioned conventional drawbacks.

[0004]

In order to achieve this object, the first characteristic configuration of the image data compression method according to the present invention is such that the image to be compressed of one frame or one field is n ×
It is divided into blocks of n (n is an integer) pixels, the gradient in a predetermined direction is obtained for the pixel data of each block, and the one-dimensional array data obtained by sampling in the direction in which the gradient is small It is configured to perform compression processing. A second characteristic configuration of the image data compression method according to the present invention is that an image to be compressed of one frame or one field is divided into blocks of n × n (n is an integer) pixels, and the pixel data of each block is composed. An image data compression method for performing compression processing on one-dimensional array data obtained by sampling in a predetermined direction, wherein the sampling direction is set for each block so that the sampling directions are different between adjacent blocks. The point is that the sampling direction of each block is changed so that the sampling direction of each block is different for each frame or field.

[0005]

According to the first characteristic configuration, the gradient in the predetermined direction is obtained in advance for the pixel data of each block, and the direction in which the gradient becomes small can be confirmed as the direction in which the pixel data of the block has a strong correlation. The one-dimensional array data obtained by sampling in such a strong correlation direction is subjected to compression processing. The gradient with respect to the constituent pixel data of the block can be easily obtained, for example, by applying a first-order differential operator used when obtaining the contour of the image or the like. According to the second characteristic configuration, since the sampling direction is made different for each block, a block in which a reproduced image with a high compression rate and excellent is obtained and a block in which a reproduced image with a low compression rate is deteriorated are mixed. However, since a similar image continues for several frames in a moving image in which one frame image is switched and displayed in 1/30 seconds, by changing the sampling direction of the corresponding block for each frame,
When the reproduced image is displayed on the display device having a long afterglow property such as a CRT, the distortion of the reproduced image can be reduced as a whole.

[0006]

Therefore, according to the first characteristic configuration, the compression processing is performed on the one-dimensional array data sampled along the direction in which the pixel data has a strong correlation for each block. It has become possible to obtain an image data compression method capable of obtaining an excellent reproduced image by performing compression with a high compression rate for the entire frame or field and with little loss of data. Further, according to the second characteristic configuration, the correlation of the pixel data in the block is preliminarily correlated with a moving image that requires high-speed processing such as switching and displaying one frame image in 1/30 seconds. It has become possible to obtain an image data compression method that can achieve improvement in the image quality of a reproduced image as a whole of continuous frames or fields without performing time-consuming processing such as checking.

[0007]

EXAMPLE A first example of the image data compression method according to the present invention will be described below. As shown in FIGS. 1 and 2, one frame of image data composed of M rows and N columns of pixels stored in the bitmap memory 1 is divided into a plurality of blocks composed of n × n pixels. , Is output to the block memory 2 for each block. The block size is
The maximum can be set around n = 16, which is the range where the correlation can be found in the constituent pixels, and this time, n = 8 is set. For example, a primary differential operator as shown in FIGS. 3A and 3B is applied to the data of each block output to the block memory 2 to detect the direction in which the gradient of the image data is the smallest.

That is, FIG. 3a shows a first-order differential operator in the upward direction, and FIG. 3b shows a first-order differential operator in the diagonally upper right direction, and each operator is multiplied by each point of the constituent pixels of the block. The differential value is obtained, and the direction in which the value is smallest is determined to be the direction in which the pixel data has a strong correlation. Specifically, as shown in FIG.
By rotating respectively, the first-order differential operator 3 obtained not only in the upward direction and the diagonal right upper direction but also in the right direction and the diagonal lower right direction is made to act, and the direction having the smallest value among the four directions. Is determined to have a strong correlation with the pixel data.

Here, the correlation of pixel data means a direction in which the repetition of bright and dark stripes is the smallest in a monotone image, for example, and in the image data decomposed into the three primary colors of RGB each color plane (three bit map memories). Is the direction in which the repetition of light and shade stripes is the smallest, and in the mixed-color image data, the direction in which the stripes formed by the colors having different RGB color mixture ratios are the smallest is defined. To be determined.

The quantizing means 5 at the subsequent stage samples the block-constituting pixel data into one-dimensional array data in the direction in which the pixel data discriminated by the discriminating means 4 has a strong correlation. Data compression is performed by quantizing the data obtained by performing the original linear transformation, or by performing run-length coding (for example, LZW method) on the one-dimensional array data.

A second embodiment of the image data compression method according to the present invention will be described below. The color moving image is composed of a plurality of still images of 30 frames / 1 on a commercial television.
Second, more specifically, two fields obtained by dividing a still image of one frame into an even field and an odd field are displayed at 1/60 second per field. Therefore, image processing such as compression / expansion of moving images can be handled as image processing such as compression / expansion of still images for each frame or for each field. Therefore, image processing such as compression / expansion of moving images will be described below as image processing for handling color still images.

As shown in FIG. 5, the bitmap memory 1
The 1-field image data composed of pixels of M rows and N columns stored in 1 is divided into a plurality of blocks composed of n × n pixels, and output to the block memory 12 for each block. The quantizer 14 obtains the one-dimensional array data by sampling the block-constituting pixel data in accordance with the sampling pattern designated by the sampling pattern table 13 and subjecting the one-dimensional array data to the one-dimensional linear conversion. Data is quantized, or run length coding (for example, LZW) is performed on one-dimensional array data.
Data compression is performed by applying the method).

The sampling patterns designated by the sampling pattern table 13 are, as shown in FIG. 4, five patterns in the downward direction, the oblique lower right direction, the oblique vertical direction, the right direction, and the oblique upper right direction between adjacent blocks. Is set so that the sampling direction of each block is different, and further, the sampling direction in each block is changed so as to be different for each field. For example, it is conceivable to shift the sampling pattern for each block in the row direction or the column direction at intervals of one block or several blocks each time the fields differ.
The method is not particularly limited.

Another embodiment of the present invention will be described below. Although the first and second embodiments do not describe the means for reproducing a compressed image, the compression method of compressed data, data indicating the sampling direction, and encoding table data, for example, in run length encoding is used. It can be decrypted by a known method based on the identifiable data.

The configuration of the circuit block described in the above embodiment is not particularly limited and can be arbitrarily configured by using a known technique. Further, the first-order differential operator is not limited to the one shown in the above embodiment.

It should be noted that reference numerals are given in the claims for convenience of comparison with the drawings, but the present invention is not limited to the configurations of the accompanying drawings by the entry.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of an image compression device.

[Fig. 2] Block diagram of primary differential operator

FIG. 3 is an explanatory diagram of bitmap data and block division.

FIG. 4 is an explanatory diagram showing a sampling pattern.

FIG. 5 is a block configuration diagram of an image compression apparatus showing another embodiment.

[Explanation of symbols]

 1 Bitmap memory 2 Block memory 3 Primary differential operator 4 Discrimination means 5 Quantization means

Claims (2)

[Claims] 1. An image to be compressed of one frame or one field is divided into blocks of n × n (n is an integer) pixels,
An image data compression method in which a gradient in a predetermined direction is obtained for the pixel data of each block, and compression processing is performed on the one-dimensional array data obtained by sampling in the direction in which the gradient is small. 2. A compression target image of one frame or one field is divided into blocks of n × n (n is an integer) pixels,
An image data compression method that performs compression processing on one-dimensional array data obtained by sampling the constituent pixel data of each block in a predetermined direction, such that adjacent blocks have different sampling directions. While setting the sampling direction for each block,
An image data compression method for changing the sampling direction of each block so that the sampling direction of each block is different for each frame or field.