JPS60248076A - Method and apparatus for coding - Google Patents

Abstract

PURPOSE:To decrease the deterioration of picture quality and to improve the compression rate by utilizing a visual characteristic of a dark part and a light part of a picture so as to divide an amplitude of a picture signal into plural luminance areas thereby deciding a constant at each luminance area. CONSTITUTION:In dividing a gradation picture subject to multi-value quantization into blocks each comprising a prescribed number of adjacent pictures and coding the picture at each block, an original picture element signal 260 is selected and inputted to an adder/divider 203, stored in one block memory 205, a mean value E0 is calculated and outputted as the mean value 263 of the picture elements in the block. The mean value E0 (signal 262) is inputted also to a constant memory 206 and a corresponding constant Cj is outputted. Then the mean value E0 of luminance is divided into plural luminance areas and a constant is decided at each luminance area, then a picture with less picture quality deterioration is obtained and the compression rate is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a facsimile machine for transmitting images with gradations such as photographs, and an image encoding method and apparatus for an image storage apparatus for storing images. It is related to.

Configuration of conventional example and its problems The document on the block encoding method shown below is an image encoding method that divides a quantized multi-gradation image into blocks consisting of adjacent pixels and encodes the blocks [:1) ,
[2), (1) has been proposed.

[1] O.R. Mitchell et al.” Block
Truncation Coding; A New
Approach to ImageCompress
ion”, IEEE International
Conference on Communicati
ons, vol.

978 [2] Kishimoto et al. “Study of block coding method for still images”, IEICE study group material IE78-43.19
78 [3] Kobayashi “Differential block coding method for multi-gradation facsimile signals”, Transactions of the Institute of Electronics and Communication Engineers, vol. 16
6 B, 42, 1983, 1] and [2]
] is an encoding method that divides the pixels in a block into two groups using the average value of the luminance of the pixels in the block as a threshold, determines the luminance for each division, and expresses the block in two gradations. The method described in Reference [3] sets a constant for the difference between the average value and the brightness of each pixel in order to further increase the compression ratio, and if the difference in brightness is within the range of the constant, the Only the average value.

In other words, a block is represented by one gradation, and if the difference in brightness is outside the constant range, each pixel in the block is expressed as 1 or more than the average value.
This is an encoding method in which blocks are divided into two groups of less than or equal to 1 and expressed by the average value of each group, that is, by two gradations. In this method, increasing the constant related to the brightness difference increases the compression ratio, but also increases the deterioration of image quality in areas where the brightness is low, that is, the optical density is high.

On the other hand, if the constant is made smaller, the deterioration in image quality will be reduced, but
The improvement rate of the compression ratio with respect to the contents described in documents [:1] and (2) is small.

Purpose of the Invention In view of the above-mentioned drawbacks, the present invention has been developed to provide a method for use in cases where a large number of gradations are required, such as in commercial photographic transmission, and in which there is little deterioration in the image quality of the reproduced image, and furthermore, a high compression ratio can be obtained. The present invention provides a method and apparatus for encoding a toned image.

Structure of the Invention In order to achieve this object, the gradation image encoding method and device of the present invention divides a multi-level quantized gradation image into blocks each consisting of a fixed number of adjacent pixels, and When encoding each image, the amplitude of the image signal is divided into multiple luminance regions, a constant is determined in advance for each luminance register, and the difference between the average luminance value of the block and the luminance signal of each pixel in the block is calculated. are all within the range of constants determined for the luminance range to which the average value belongs, the average value and a 1-bit sign indicating that this block is represented only by the average value are used as block information, and the average value and If there is at least one pixel in which the luminance signal difference of each pixel is outside the range of a constant determined in the luminance region to which the average value belongs, a pixel with a luminance signal greater than or equal to the average value and a pixel with a luminance signal less than the average value The pixels in the block are divided into two parts, and each division is given the average value of the pixels to which it belongs, a code indicating which average value each pixel belongs to, and a code indicating that this block has two types of average values. Each block is encoded using the 1-bit code shown as block information.

The image signal obtained when a gradation image is scanned using optical means has a value proportional to the brightness of the document1, and the difference in brightness signal between pixels in the dark part of the image, that is, the part with high optical density. Even if the difference in luminance signals between pixels in the bright part of the image, that is, the part with low optical density, is the same, the optical density difference on the pixel surface is significantly different between the dark part and the bright part. This is no good,
A block that encodes only the average value with block encoding,
The pixels in a block are divided into two groups, each having an average value or more and less than 2, and each average value is encoded into a block. It is determined by the average value of the brightness of pixels, that is, the average value of brightness is divided into multiple brightness regions,
By setting a constant for each luminance recording, an image with less deterioration in image quality can be obtained in both dark and bright areas, and the compression ratio can be increased.

DESCRIPTION OF EMBODIMENTS An encoding device according to an embodiment of the present invention will be described below with reference to the drawings.

Note that FIG. 1 shows a block diagram of an embodiment of a system used in the present invention. In FIG. 1, reference numeral 1 denotes an image reading device, which scans, samples, and quantizes gradation images such as photographs. 2 is a device for encoding a multi-level quantized image.

Reference numeral 3 designates a transmitting device that transmits the code, 4 a receiving device that receives the transmitted code, 6 a decoding device that reproduces an image from the code, and 6 an image recording device that records the reproduced image.

Hereinafter, the encoding device 2 and the decoding device 6 will be described with reference to FIG.
The configuration will be explained in more detail.

FIG. 2 is a block diagram showing a more detailed configuration of the encoding device 2. As shown in FIG. The quantized image signal 8260 scanned by the image reading device 1 ) is stored in the buffer 1 memory 201 . Note that this encoding device 2 divides an image into blocks each having a fixed number of adjacent pixels, and performs encoding processing on a block-by-block basis. In the following description, a block is assumed to be a rectangle consisting of m pixels in the main scanning direction and n pixels in the sub-scanning direction, and m x n = N.

Now, a signal is read out from the buffer memory 201 one pixel at a time in the middle of the block (original pixel signal 260). The original pixel signal 260 is selected by the selector 202 and input to the adder/divider 203, and at the same time it is input to the memory 2 for one block.
I have a huge admiration for 05.

Then, the average value E0 (signal 2
62) is calculated by the following equation, selected by the selector 204, and output as the average value 263 of the pixels in the block.

(However, xi is the luminance signal value of the pixel) On the other hand, the average value E. is also input to the constant memory 206, and the average value E is input from the constant memory 206. The constant G corresponding to (
signal 265). The configuration of constant memo IJ 206 is shown in the table. The table is an example where the average value luminance range is divided into four.

Table average value E. After completing the calculation, the pixel values (signal 264) in the block are sequentially read out from the memory 206 for one block, and the average value E is obtained by the subtracter 20. At the same time, the comparator 210 compares the value of each pixel with the average value E0 and outputs the absolute value of the difference (signal 267).
0 is output to the shift register 211 and selector 212. The comparison result is the average value E. 0 if above, average value 8
If it is less than 8, 1 is assigned. The value of the pixel in the block (signal 264) is the average value E according to the comparison result 270. The first of the above is rape (signal 271)
and a second group (signal 272) with an average value of less than 80, and are stored in memories 213 and 214, respectively.

Each pixel in the block and the average value E. absolute value of the difference (signal 2
67) is compared by the comparator 208 with a constant Cj defined in the luminance range to which the average value Eo belongs, and a comparison result 268 is output to the determiner 209.

After comparing all the pixels of one block, the determiner 209 selects the block identification signal 269 to 1 if there is at least one pixel equal to or greater than the constant G. When the block identification signal 269 is 0, that is, if the difference from the average value E0 for all pixels in the block is within a constant C1, the code 216 indicates the average value E0 and a 1-bit block identification signal (for example, value 0). This block to be encoded is called S-pro. On the other hand, when the identification signal 269 is 1, the first group is read out from the memory 213 via the selector 202, and the adder/divider 203 calculates the average value E1 (signal 264) of the first group and the second group. memory 214
The average value E2 (sign'j) 265) of the second group stored in is calculated. That is, is calculated. Here, 11 * K1 is the value and number of pixels belonging to the first group, Zi and K2 are the value and number of pixels belonging to the second group, and K is 2 to 10.
=N.

When the block identification signal 269 has a value of 1, the encoder 216 uses the average value E and K2, N bits of group attribute information of each pixel stored in the shift register 211, and 1 bit of block identification information (for example, a value of 1). It is encoded and output (signal 276).

Note that this block is called a D block.

When the encoder 215 directly outputs the information to be encoded as the code output 276, if one image is quantized to 8 pits, in the case of an S block, the number of code bits per pixel includes the block identification information and the identification Including information-(17
+N) bits/pixel N.

The image encoded by the above device has the value E for S blocks. In the case of D block, the group attribute information is used to generate N pixels with the value E1 for the first group of pixels and K2 of the second group of pixels.
It is reproduced by generating pixels with .

As described above, according to this embodiment, the number of encoded bits per pixel of an 8-bit quantized image is set when the probability of occurrence of an S block is P, and the higher the probability of occurrence of an S block, the higher the compression ratio. The probability of occurrence of an S block increases as the constant Cj increases. However, when the constant Cj is large, image quality deterioration in dark areas (portions with high optical density) of the image increases. In the case of this embodiment, the constant C differs depending on the average value E0, so C1 is made small in areas with high optical density, that is, areas with small brightness values, and C1 is made small in areas with low optical density. By increasing C), it is possible to obtain an image with less deterioration in visual quality in both dark and bright areas.

Effects of the Invention As described above, the present invention utilizes the visual characteristics of dark areas, that is, areas with high optical density, and bright areas, that is, areas that have low optical density, to adjust the amplitude (brightness) of an image signal to a plurality of values. By dividing the image into brightness ranges and setting a constant for each brightness recording, it is possible to reduce the visual deterioration of image quality and increase the compression rate, which has great industrial value.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an embodiment of an image transmission system used in the present invention, and FIG. 2 is a block diagram of an encoding device in an embodiment of the present invention. 1... Image reading device, 2... Encoding device, 3... Code transmitting device, 4... Code receiving device, 5......・Decoding device, 6...
Image recording device, 201...B. Family memory, 202.204.212...Selector, 203...Adder/divider, 205, 21
3.214...Par memory, 206...Constant memory, 207...Subtractor, 208,210...
...Comparator, 209・River...Determiner, 211...
・・・Shift register, 216・old・・encoder. Name of agent: Patent attorney Toshio Nakao, and one other person Procedural amendment document: November 1988, Director General of the Japan Patent Office 2. Name encoding method of invention and its device 3. Person making the amendment Relationship with the case Patent issue Request Address: 1006 Kadoma, Kadoma City, Osaka Prefecture Name (
582) Matsushita Electric Industrial Co., Ltd. Representative Toshihiko Yamashita 4 Agent 571 Address 1006 Oaza Kadoma, Kadoma City, Osaka Matsushita Electric Industrial Co., Ltd. Claims column 6 of the specification subject to amendment 5, amendment Contents (1) The claims column of the specification will be amended as shown in the attached sheet. / (- Change "code 216" on the 4th line of the 1st page to "encoder 21
5". (3. On page 14, line 14 of the same, "code transmitting device, 4... code receiving device" is amended to "transmitting device, 4... receiving device". 2. Scope of Claims (1) ) A multilevel quantized gradation image is divided into blocks consisting of a fixed number of adjacent pixels, and the amplitude of the image signal is divided into a plurality of luminance regions, a constant is determined in advance for each luminance register, and the above-mentioned If the difference between the average luminance value of a block and the luminance signal of each pixel in the block is all within the range of a constant determined in the luminance range to which the average value belongs, the average value and this block are expressed only by the average value. If there is at least one pixel whose luminance signal difference between the average value and each pixel is outside the range of a constant determined in the luminance region to which the average value belongs,
dividing the pixels in the block into pixels having a luminance signal greater than (or exceeding) the average value and pixels having a luminance signal less than (or below) the average value, and determining the average value of the pixels belonging to the valley division;
Each block is encoded using a code indicating which average value each pixel belongs to and a 1-bit code indicating that this block has two types of average values as block information. encoding method. A first storage means for dividing and storing a multilevel quantized gradation image into blocks each consisting of a certain number of adjacent pixels, and a block average value calculation means for calculating the average value of the brightness of the blocks. and constant assigning means for dividing the signal amplitude of the gradation pixel into a plurality of luminance ranges and assigning a constant corresponding to the average value to each luminance register; determining means for determining whether or not all the differences between the luminance signals of each pixel are within a constant range defined in the luminance region to which the average value belongs; block dividing means for dividing into a block of pixels having a luminance equal to or greater than the average value and a block of pixels having a luminance less than (or below) the average value; Indicates each average value of a pixel block having a luminance greater than (or exceeding) the average value and a pixel block having a luminance less than (or below) the average value, and which average value each pixel belongs to. A device for converting a code into a 1-bit code indicating that the block has two types of average values.

Claims (2)

[Claims] (1) A multilevel quantized gradation image is divided into blocks consisting of a fixed number of adjacent pixels, and the amplitude of the image signal is divided into multiple luminance regions, and a constant is determined in advance for each luminance register. , if the difference between the average luminance value of the block and the luminance signal of each pixel in the block is all within a constant range defined in the luminance range to which the average value belongs, then the average value and this block are only average values. A 1-bit code indicating that a pixel is displayed is used as block information, and there exists at least one pixel for which the luminance signal difference between the average value and each pixel is outside the range of a constant determined in the luminance region to which the average value belongs. Ba,
dividing the pixels in the block into pixels having a luminance signal greater than (or exceeding) the average value and pixels having a luminance signal less than (or below) the average value, and determining the average value of the pixels belonging to each division;
Each block is encoded using a code indicating which average value each pixel belongs to and a 1-focus code indicating that this block has two types of average values as block information. encoding method. (2) A first storage means that divides and stores a multilevel quantized gradation image into blocks each consisting of a certain number of adjacent pixels, and a block average value calculation that calculates the average luminance value of the blocks. means for dividing the signal amplitude of the gradation pixel into a plurality of luminance regions and assigning a constant corresponding to the average value to each luminance register; determining means for determining whether or not all the differences between the luminance signals of each pixel are within a constant range defined in the luminance region to which the average value belongs; and when the determining means determines that the difference is outside the range, the block block dividing means for dividing into a block of pixels having a luminance equal to or greater than the average value and a block of pixels having a luminance less than (or below) the average value; Indicates each average value of a block of pixels having a luminance equal to or greater than (or exceeding) the average value, and a pixel block having a luminance less than (or below) the average value, and which average value each pixel belongs to. sign and
An encoding device comprising divided block processing means for assigning a 1-bit code indicating that the block has two types of average values.