JPS6268369A - Picture coding system - Google Patents

Abstract

PURPOSE:To obtain a picture with excellent S/N and high quality by applying coding in the unit of a group larger than a block with respect to non multi- gradation picture information to improve the coding efficiency. CONSTITUTION:Input picture information is split into plural blocks comprising plural picture elements to obtain two (large and small) block representative gradations an, bn representing the gradation of a picture element in the block at each block. Then the two (large/small) group representative gradations A, B representing the gradation of the picture element in the group are decided for the group using consecutive >=2 blocks as one unit. The block representative gradations an, bn and the group representative gradations A, B are used to decide whether or not the input picture information is non multi-gradation picture information, and the input picture information decided to be non multi- gradation picture information encodes the gradation information and resolution information in the unit of groups and when not, that is, the multi-gradation picture information is coded in the unit of blocks in the same way as a conventional block coding system.

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a high-efficiency compression encoding method for image information, and in particular? The present invention relates to an image encoding method suitable for a combination of value images and multi-gradation images.

[Technical disadvantages of the invention and its problems]

Among the high-efficiency compression encoding methods for image information used in facsimiles, etc., the block encoding method is
It is attracting attention as an excellent image encoding method that can handle both value images and multi-tone images such as photographs.

The principle of the block encoding method will be explained using the second F.

As shown in FIG. 2(a), the original image information is
It is divided into blocks of pixels, and a threshold value xT, which is the average gradation of the pixels in the block, is determined. This threshold value X7 and the gradation (luminance level) of each pixel X + (i-1, 2, -
, 9), Xl is larger than xT (X+
≧XT), when φi-1°×1 is smaller, it is determined as φi-o. Then, the average gradation of pixels whose gradation is smaller than the threshold value X7 is defined as the average gradation of pixels whose gradation is larger than the threshold value X7. A and b are called block representative gradations, and the gradation Xi of each pixel in the block is replaced by these block/representative gradations a and b to become Xi'. Also, φH(i=1.2
．． ..., 9) is called resolution information.

These representative gradation information a, b and resolution information φ1 are transmitted using an appropriate encoding method. FIG. 2(b) shows a specific numerical example of FIG. 2(a). In addition,
In the example shown in FIG. 2(b), the average gradation of pixels within the block is used as the threshold value×T.

However, this block encoding method does not allow continuous changes in gradation in areas of non-multi-gradation images, such as binary images such as documents or background areas of photographs. Since the representative gradation information is changed for each unit for encoding, there is a problem in that the encoding efficiency decreases. In addition, increasing the block size improves encoding efficiency, but because the gradation changes greatly at the boundaries between blocks, image discontinuity between blocks, that is, block structure, tends to appear, reducing image quality. do.

In order to solve the problems of such ordinary block coding methods, we have published the IEICE Study Group Material r E 81-5゜19.
As described in 81, 9.33 to 40, etc., when the difference between two large and small block representative gradations a and b is small, or when the number of pixels Na and Nb corresponding to each representative gradation is larger than a certain value. integrates the representative gradation information into one, and converts the resolution information φ1 into MH (Modified Huffman code), MR (
A differential block encoding method has been proposed in which the representative gradation is encoded using a Weyl code or the like, and its run length is encoded using a Weyl code or the like.

According to this differential block coding method, although the coding efficiency is improved, in the area of non-multi-tone images, only one type of representative tone information is required. There is a waste of change. Furthermore, since the representative gradation information is run-length encoded, if even one gradation changes between adjacent blocks, the run is interrupted, resulting in a decrease in encoding efficiency. Furthermore, although the block structure is reduced on average in the decoded and reproduced image, the block structure is more noticeable in areas where the gradation changes are gradual, such as in the discreet parts of a photograph.

[Purpose of the invention]

The present invention has been made in order to solve the above-mentioned conventional problems, and it is an object of the present invention to provide an image encoding method that has high encoding efficiency and can suppress the occurrence of block structures that cause deterioration of reproduced images. shall be.

(Summary of the Invention) In order to achieve this object, the present invention divides input image information into a plurality of blocks each consisting of the same number of pixels, and each block represents the lJ tone of the pixels in that block. Two large and small block representative gradations an, bn (
n=1.2. -), and determine two group representative gradations A and B (A<B) that represent the gradation of pixels in the group for a group consisting of two or more consecutive blocks as one unit. , these block representative gradations a
rL.

It is determined whether the input image information is non-multi-gradation image information using bn and group representative gradations A and B, and input image information determined to be non-multi-gradation image information is divided into gradation information and resolution information. is encoded in groups, and if not,
In other words, multi-gradation image information is characterized in that it is encoded in units of blocks, similar to the conventional block encoding method.

In other words, the present invention does not always encode in block units as in conventional block encoding or differential block encoding methods, but when the input image information is a binary image such as a document image or the background part of a photograph. For image information with almost no gradation changes (this is called non-multi-gradation image information), two or more blocks that are considered to be non-multi-gradation image information are treated as one group, and the information is divided into groups. It encodes gradation information and resolution information.

〔Effect of the invention〕

According to the present invention, non-multi-gradation image information is encoded in units of groups larger than blocks.
Encoding efficiency is improved.

In addition, in such a non-multi-gradation image area, constant gradation information is transmitted over the entire area, so the phenomenon of block structure due to gradation changes between blocks in the reproduced image is eliminated, and S/ N good high quality images can be obtained.

[Embodiments of the invention]

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a functional block diagram for explaining an overview of an image encoding method according to an embodiment of the present invention. The input image information is the threshold Il! ! A threshold value XTn (n=1.2.
) is set. This threshold 11Xrn is above the average gradation mark of the pixels in the block
・Old...(3)rjfu1 It can be determined by: Further, the threshold lIX T n may be set using iterative calculation so that the mean square error is minimized.

The threshold value xTn thus set is given to the block representative tone calculation means 2 together with the input image information. The block representative gradation calculation means 2 compares the threshold XTn with the gradation (Ir 11 degree level) Xi of the pixels in each block,
Two block representative gradations aA and bn, large and small, are calculated from equations (1) and (2), and resolution information φi is output at the same time.

The block representative gradations man and bn are given to the group representative gradation determining means 3. The group representative gradation determining means 3 determines a group whose minimum unit is two or more consecutive blocks, and selects two group representative gradations A, large and small, representing the gradations of pixels within the group.

Determine B. The group representative gradations A and B are as follows:
For example, the block representative MilarL in the group.

The first one (n-1 one) or the smallest one of bn can be adopted.

Block representative gradation an, b11 and group representative gradation A,
8 is given to the determining means 4, and it is determined whether the input image information is non-multi-gradation image information. For example, this judgment is as follows:
Conditions ■～■ ■ B≦afi <B+vy, and B≦bn<B+w■
A≦afi<A+w, and A≦brL<A+w■
A≦an<A+w, and B≦l) A<B+w (however,
W is a minute gradation level difference), and when at least one of these conditions is satisfied, it is determined that the image information is non-multi-gradation image information (
(See Figure 3). Here, non-multi-gradation image information refers to images other than multi-gradation images such as photographic images, that is, 211! images such as documents, as described above. Refers to information about almost single-tone images, such as the background part of an image or photograph.

Among the three conditions (■) to (■) above, condition (2) is when the block representative gradations arL and bn in the group are both within the gradation level area 31 on the upper side of FIG. , bn are both included in the lower gradation level area 32 in FIG. The block representative floor y4brL of the higher level is the third
Gradation level area 1ii on the upper side of the diagram! 31, respectively.

The determination result of the determination means 4 is given to the encoding means 5. The encoding means 5 inputs the block representative gradations an, b, 1 and the resolution information φi, and if the determination result of the determining means 4 is non-multi-gradation image information, the encoding means 5 inputs the representative gradation information and resolution information in group units. If the image information is not multi-gradation image information, representative gradation information and resolution information are encoded in units of blocks as in the conventional block encoding method. As for the format of this encoding, for example, the representative gradation information is represented by the representative gradation arL as shown in FIG. 4(a).

The information representing the level of brL and its run length are encoded, and the resolution information φ1 is shown in the same figure (b).
As shown in , the pixel has two representative gradations an and b, large and small.
Following the 1-bit information indicating which group n belongs to,
The run length can be encoded.

Also, among the above conditions ■ to ■, only condition ■ or condition ■
If only the following holds true, similar to the conventional differential block encoding method, when encoding the representative gradation information, the two large and small representative gradation information an and bn may be combined into one and encoded. That is, the conventional differential encoding method and the group-based encoding based on the present invention are used together. By doing so, it is possible to further improve the encoding efficiency.

The image encoding method according to the present invention described above can also be executed by software processing using a microprocessor. The embodiment will be explained with reference to the flowchart shown in FIG. In FIG. 5, at step 11, one block of input image information is taken in, and at the same time, n=1, and at step 12, a threshold value X is set. Next, in step 13, the block representative gradation an is determined using the threshold value x digit n.

bn is calculated, and then in step 14 the group representative floor 5 is calculated.
Set 1A and B.

Next, in step 15, n=n+1 is set, and at the same time, in step 16, it is determined whether the end of the line, that is, the end of each main scan has been reached. If it is determined that the line has ended, the line is updated in step 17, that is, the process moves to the next main scan, and the process returns to step 11. If the determination result in step 16 is that the line is not finished, the next block of input image information is captured in step 17, and the threshold XTA is set in step 19.20 in exactly the same manner as step 12.13. Furthermore, the block representative gradation an, kg is calculated.

When step 20 is completed, the next step 21 is to calculate the block tone mill obtained so far.

Based on bA and the group representative gradations A and B, the presence or absence of the three conditions (1) to (2) is checked, and it is determined whether the image information is non-multi-rP3-tone image information. As a result, if it is determined that the image information is non-multi-tone image information, the process returns to step 15. Furthermore, if it is determined that the image information is not multi-tone image information, the process moves to the next step 22, and the representative tone information and resolution information for the image information of (n-1) blocks are encoded. The encoding in step 22 encodes the representative gradation information and resolution information for one block of image information when n=2, so it is different from the normal block encoding method (or differential block encoding method). The same is true. On the other hand, n=
If the number is 3 or more, in step 22, two or more blocks are treated as one group and the representative gradation information and resolution information within that group are encoded. That is,
Encoding according to the invention will be performed.

When the encoding process in step 22 is thus completed, the process returns to n-1 in step 23, and then, after the group representative gradations A and B are updated in step 24, the process returns to step 15 and n is incremented by one again. The processing from step 16 onwards is repeated.

As described above, according to the present invention, for image information of an area determined to be non-multi-tone image information, representative tone information and resolution information are encoded in units of groups consisting of a plurality of consecutive blocks corresponding to the area. By doing so, it is possible to improve encoding efficiency and image quality.

That is, in the present invention, as seen in the distribution of representative gradations shown in FIG. 3, the area of the binary image is black.

There is no need to encode representative gradation information for each block even in a period that includes a white change point, and the same representative gradation information is used during that period, so conventional block encoding methods and differential blocks There is no reduction in efficiency as with encoding methods.

In addition, it is possible to improve coding efficiency without increasing the block size, and it is also possible to group image information not only in binary images but also in areas with gradual gradation changes, such as the background of a photo. By performing unit encoding, it is possible to suppress the occurrence of block structures in decoded and reproduced images.

This effect is also clear from the experimental results shown in FIGS. 6 and 7. FIG. 6 shows the case where a document image and a photographic image are encoded using the conventional differential block encoding method and the block concatenation encoding method based on the present invention, respectively.
Number of bits per pixel of encoded output (number of encoded bits)
This figure shows the results of actually measuring the relationship between SDD/Nrffis (ratio of signal peak-to-peak and effective value of noise) of the reproduced image. As the photographic image, a standard image called a girl image, which is generally used in the evaluation of image processing, was used. From this result, the S Dp/N rn+s of the reproduced image is greatly improved for the same number of encoding bits, and the improvement effect is particularly remarkable in the document portion, which is a binary image. A considerable improvement can also be seen in photographic images, especially in the dorsal area (SI).
This is an effect due to an improvement in D/N r+ns. Furthermore, FIG. 7 shows the results of similar measurements for an image containing a mixture of document images and photographic images.
This figure also shows that the block concatenation coding method according to the present invention is more sophisticated than the differential block coding method.

Note that the present invention is not limited to the above embodiments,
For example, in the actual MIN, the block is a square of 3 x 3 pixels, but it may also be a rectangle, and the block size can be made larger.

In addition, in the embodiment, an example was explained in which blocks were connected in the horizontal direction (main scanning direction) and grouped, but in the vertical direction (main scanning direction)
They may be connected and grouped in the backward scanning direction). Further, in the actual example, the number of blocks constituting one group is variable, but it may be fixed to an appropriate value.

Furthermore, the present invention is applicable to videotex other than facsimile,
It can also be applied to encoding still images such as in filing, and can also be applied to encoding moving images in television signal transmission, recording, etc. In addition, the present invention can be implemented with various modifications without departing from the scope.

[Brief explanation of drawings]

FIG. 1 is a functional block diagram for explaining an image encoding method according to an embodiment of the present invention, FIG. 2 is a diagram for explaining a conventional block encoding method, and FIG. 3 is a functional block diagram for explaining an image encoding method according to an embodiment of the present invention. FIG. 4 is a diagram for explaining block concatenation encoding performed on non-multi-gradation image information in the image encoding method. FIG. 4 is a diagram showing the format of the encoded output obtained by the image encoding method of the present invention. The figure is a flowchart for explaining an embodiment in which the image encoding method according to the present invention is executed by software processing, and FIGS. 6 and 7 are a differential block encoding method and a block concatenation encoding method based on the present invention. FIG. 3 is a diagram showing the relationship between the number of encoded bits and the .S pM N ribs of a reproduced image in the case of encoding. DESCRIPTION OF SYMBOLS 1... I1m setting means, 2... Block representative gradation calculation means, 3... Group representative gradation determining means, 4...
Non-multi-gradation image determining means, 5... encoding means. Applicant's agent Ben 1'l Takehiko Suzue Figure 3 Figure 4

Claims (7)

[Claims] (1) In an image encoding method that encodes input image information by dividing it into representative gradation information representing the gradations of multiple pixels and resolution information indicating which representative gradation information each pixel corresponds to. , the input image information is divided into a plurality of blocks each composed of a plurality of pixels, and for each block two large and small block representative gradations a_n, b_n (n= 1, 2, ...), and then calculate two group representative gradations A, B( After determining A<B), it is determined whether the input image information is non-multi-gradation image information using these block representative gradations a_n, b_n and group representative gradations A, B, and the non-multi-gradation image is determined. An image characterized in that input image information determined to be information is encoded with representative gradation information and resolution information in group units, and other input image information is encoded with representative gradation information and resolution information in block units. Encoding method. (2) The block representative gradations a_n and b_n are such that the average gradation of pixels in each block is a threshold, and the average gradation and gradation of pixels whose gradations are larger than the threshold are smaller than the threshold. 2. The image encoding method according to claim 1, wherein the image encoding method is an average gradation of pixels. (3) A patent claim characterized in that the block representative gradations a_n, b_n have a minimum mean square error between the block representative gradations a_n, b_n and the gradations of pixels in each block. The image encoding method according to the first item in the range. (4) In order to determine whether the input image information is non-multi-gradation image information, the block representative gradation a_n, b_n
and using the group representative gradations A and B, the following three conditions [1] to [3] [1] B≦a_n<B+w, and B≦b_n<B+w[
2] A≦a_n<A+w, and A≦b_n<A+w[3
]A≦a_n<A+w, and B≦b_n<B+w (where w is a slight gradation level difference), and when at least one of these conditions is satisfied, it is determined that the image information is non-multi-gradation image information. An image encoding method according to claim 1. (5) In order to determine whether the input image information is non-multi-gradation image information, the block representative gradation a_n, b_n
and using the group representative gradations A and B, the following three conditions [1] to [3] [1] B≦a_n<B+w, and B≦b_n<B+w[
2] A≦a_n<A+w, and A≦b_n<A+w[3
]A≦a_n<A+w, and B≦b_n<B+w (where w is a minute gradation level difference), and when at least one of these conditions is satisfied, it is determined as non-multi-gradation image information, and further conditions are established. Claim 1, characterized in that when only [1] or only [2] holds true, when encoding the gradation information, two pieces of representative gradation information, large and small, are integrated into one and encoded. The image encoding method described in Section 1. (6) The image code according to claim 1, wherein a determination as to whether the input image information is non-multi-gradation image information is made every time block representative gradations a_n, b_n are newly given. method. (7) The first aspect of the present invention is characterized in that the determination as to whether the input image information is non-multi-gradation image information is made every time a predetermined number of block representative gradations a_n, b_n are given.
Image encoding method described in section.