JPS6238079A - Picture coding device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] Industrial Applications The present invention utilizes the correlation between neighboring pixels of image data, eliminates redundant takes, and encodes images with less code amount and less deterioration in image quality. The present invention relates to an image encoding device that performs.

BACKGROUND OF THE INVENTION Semiconductor manufacturing technology has made remarkable progress in recent years, and memories are rapidly becoming larger in capacity and lower in price.

Along with this, advanced image processing techniques using frame memories have become commonplace, and research and development of more efficient image encoding techniques has begun.

FIG. 3 shows the positioning of the image encoding and decoding apparatus when applied to an image transmission system. In Fig. 3, 301 is an image information input device that reads image information using a camera, scanner, etc., and 302 is an image encoding device that removes redundancy and compresses input image signals by using synchronization between neighboring pixels. , 303 is a transmitting device that sends the encoded image signal to a transmission path, 304 is a receiving device that receives the encoded image signal sent through the transmission path, and 305
306 is an image encoding device that reproduces an image signal from the encoded image signal, and 306 is an image display device that displays the reproduced image.

Conventionally, as a high-efficiency encoding method for multi-gradation images, the image is divided into appropriately thick blocks that group neighboring pixels, and redundancy is reduced using pixel correlation within the block or correlation between neighboring pixels. There are ways to suppress and encode it. An example of this method is ``Kishimoto, Mikuya et al.: ``Study of block coding method for still images'', IEICE Technical Report IE78-43JJ.
There is.

In this method, several appropriate gradations are assigned to each block, and the luminance values of each pixel are represented by these gradations. The more gradations you assign, the more faithfully the original image can be reproduced, but the number of encoding bits increases, so usually two gradations are set for each block, and each pixel is By specifying which key the code belongs to, the number of encoding bits is reduced while still maintaining precision. The general formula for the block encoding method in this case is expressed by equation (1).

1st pixel x in Pl block: Luminance level Yl of pixel Pi: Luminance level ao after conversion by encoding of pixel Pi
, al Two tone levels φ assigned within the niblock
l: A binary number indicating whether the luminance of pixel P1 is 80 or 81.
The threshold value in a block, that is, the brightness level, is expressed by only two gradations of O+a+, and which brightness level is assigned to each pixel can be expressed by 1 bit, φl is 0 or 1. An advantage of this method is that the number of bits per pixel can be significantly reduced compared to when the brightness of each pixel is independently encoded.A conceptual diagram of a block encoding table is shown in FIG.

In the figure, 401 divides the image horizontally and vertically,
Blocked state i, 402 is a pixel value in a certain block, 403 is ao assigned to the block,
Pit pattern of each pixel for two gradations of al, 404
405 represents the pixel value after the decoding conversion process, 405 represents the encoding conversion process, and 406 represents the id decoding conversion process.

Here, if the two gradations of 'O1a+ are expressed with q bits each and φl with 1 bit each, the bits per pixel [B
is expressed by equation (4).

B=1+2CI/N (4) However, when the original image is expressed with 8 bits per pixel, approximately nq=8 is required to obtain sufficient image quality, and N = 4.
When encoding in block units that fit from X4 pixels,
Compressed to 2 bits per pixel.

However, with the above configuration, if you try to further reduce the pixel per pixel, you will have to increase the resolution by increasing the block size or limiting the number of gradations. There have been problems in that gradation is impaired and image quality is inevitably degraded. The present invention aims to solve the above problems and achieve even more efficient encoding.

Means for Solving the Problems The present invention blocks an input image by grouping neighboring pixels into one unit, and gives a plurality of gradations to the block based on the values of pixels in the block, Provide pattern information indicating which gradation is assigned to each pixel in the block, encode each gradation when creating a code by combining the gradation information and pattern information, and encode those gradations. By changing part of the pattern information depending on the order in which the pattern information is used, the number of codes in the pattern information can be reduced and the total number of codes can be reduced.

According to the above configuration, the present invention actively utilizes the information on the magnitude of the gradation value, and replaces a part of the button information with the information on the order when encoding the gradation information, thereby generating a pattern. The number of codes for information is reduced, and overall the number of codes can be reduced without any loss in image quality compared to the conventional method.

Example An embodiment of the present invention will be described below.

FIG. 1 is a block diagram of an image encoding device according to an embodiment of the present invention. In FIG. 1, 101 is an image input unit that takes in image information as digital data;
An image storage section 103 stores the image information converted into two digits; 103 is a blocked data buffer section that reads the data of adjacent pixels from the image storage section 102 into blocks and temporarily stores them; A gradation separation unit that separates and extracts binary gradation information from image information, 1
05 is a gradation information storage unit that temporarily stores separated gradation information; 106 is a pattern information storage unit that temporarily stores pattern information necessary for associating pixels in the Glock with gradation information; 107 is a gradation information encoding order determining unit for replacing the first correspondence between a specific pixel and gradation information obtained from the pattern information into the relationship between the encoding orders of two gradation information; 108 r/'i gradation; Information encoding order determining unit 107,
Gradation information storage section 105 and pattern information storage section 106
This is a code creation unit that synthesizes the data sent from the computer, creates a code, and outputs it.

The operation of the above configuration will be explained below.

In the image input section 101, after reading the gradation image with a TV camera, sampling is performed at the pixel period, quantized at 8 bits per pixel, and then sequentially transferred to the image storage section 102. The image storage unit 102 stores digitized information for each pixel. Blocked data buffer section 1
03, the value of each pixel is read out from the image storage unit 102 for each block consisting of 4 pixels in the horizontal direction and 4 pixels in the vertical direction (total number of pixels N=16 pixels) and temporarily stored as a unit of encoding. . In the tone separation section 104,
The gradation information in the program is expressed using two gradations representing the block, and at the same time, pattern information is created to indicate pixels corresponding to each gradation. At this point, gradation representation is performed using 8 bits each. Two floors gN a 0. δ
The ice formation method in No. 1 is based on equations (5) and (6).

8, -1 da, 1.1 4j=l (1 x1 Luminance level of pixel P1 in niblock X□Average luminance level of pixels in niblock 16: Xi < xT
The number of pixels is 11:xl>xT,,! :Number of pixels: (5L) Here, the two gradation information "0+"l obtained from equation (6) are stored in the gradation information storage unit 105, and the number of pixels obtained from equations (7) and (8) is stored in the gradation information storage unit 105. Pattern information φi (i = 1
．． 2, . . . , 16) are stored in the pattern information storage section 106 as 16-bit data.

The gradation information encoding order determination unit 107 determines the gradation information encoding order as follows based on the value φ16 of the least significant bit of the pattern information.

i) When φ16=φ, encode 1 in the order of "Or 'l"
1) When φ16-1, al and ao are encoded in this order. Here,
In the case of this embodiment, since the luminance value within the block is binary, the value of φ16 is transferred to the code generation unit 108 as is. The code creation unit 108 receives the transfer data 'O+' l rφ1(
A code is created using i=1.2, ... song, 16).

The code structure at this time will be explained with reference to FIG.

FIG. 2 is a diagram showing the structure of code data created by the code creation unit 108. In FIG. 2, (a) shows the structure of code data when φ16=0, and (b) shows the structure of code data when i-tφ16=1. 201 represents a low brightness level, 202 represents a high brightness level, 203 represents pattern information that is encoded, and 204 represents pattern information that is not encoded. When φ1620, the configuration is as shown in Figure 2 (a), and when φ16 = 1, the configuration is as shown in Figure 2 (b), and the created code is 201
．． They are output in the order of 202 and 203, and the part 204 is not output. Bit allocation is gradation information a. 8 bits each in +al, pattern information φ1 (i=1.2,...,N-1
) is 15 bits, resulting in a total of 31 bits, and in this embodiment, the code amount is reduced by 1/32 x 100 = 3.1% per bit. In addition, in the decoding device, the transferred data are set as bo, bl, φ1 (i=1.2,..., 15) in order, and k) is set as O, based on the magnitude relationship a. +al is determined,
' By determining φ16 from the order of Or al and further substituting these values into equation (1), a reproduced image is obtained.

Incidentally, in FIG. 1, the processing in each section from 103 to 108 is repeatedly performed for all or part of both input image areas.

Effects of the Invention As described above, the present invention provides a plurality of gradations to a Glock consisting of adjacent pixels, creates all the pattern information Lz of pixels corresponding to each gradation, and creates two pieces of information including these gradation information and pattern information. When encoding image information block by block, the number of codes for pattern information can be reduced by replacing part of the pattern information with ranking information during gradation information encoding, which was not used at this scale. Overall, the number of codes can be reduced without any loss in image quality compared to conventional methods.
In addition, since the number of codes per block is constant, it is possible to preserve the original feature of block coding, which is that transmission errors are less likely to spread to other blocks. Furthermore, in the present invention, the number of codes related to pattern information is 13r! In fact, one bit can be reduced, so the smaller the block size and the smaller the number of quantization bits, the greater the effect.

[Brief explanation of the drawing]

FIG. 1 is a Glock connection diagram of an image encoding device according to an embodiment of the present invention, and FIG. 2 is a conceptual diagram showing the bit allocation of a code take created by a code creation unit, which is a main part of the device.
Figure 3 is a block diagram of a conventional image encoding device;
The figure is a conceptual diagram illustrating an example of a conventional encoding method. 101... Image input unit, 102... Image storage unit, 103... Blocked data buffer unit, 104... Gradation separation unit, 105...・・・
Gradation information storage unit, 106...Pattern information storage unit, 107...Gradation information encoding order determination unit, 1
08... Code creation section. Name of agent: Patent attorney Kazuo Nakao and 1 other person No. 4
Figure 6

Claims (1)

[Claims] Blocking means for extracting neighboring pixels together as one unit from an input multi-tone image, and gradation assignment means for allocating a plurality of gradations to each unit based on the values of pixels within each unit. a pattern information creating means for creating a pattern by making a correspondence between the plurality of assigned gradations and each pixel in a unit; at least an encoding order determining means for replacing the information with information on the order in which the encoding order is to be encoded; and a code creating means for encoding a part of the gradation and pattern information excluding part of the information according to the order determined by the encoding order determining means. An image encoding device having: