JP2825697B2 - Image coding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image encoding method for encoding a multi-valued image, such as a photograph, in a facsimile or the like.

[0002]

2. Description of the Related Art Conventionally, techniques in such a field include:
For example, there is one described in the following literature. Reference 1: Proceeding of the SID, 17/2 (197
6) (US) R. Floyd et.al, “An Adaptive Algorithm For Special Grayscale
(An Adaptive Algorithm for Spatial Grayscale) "Reference 2; Kibukibuki," Signal Processing of Images for FAX and OA "(57-10-20), Nikkan Kogyo Shimbun, P.12
5-127 As described in Reference 1, the multi-valued image is referred to as “white”.
An error diffusion method is one of the methods of expressing the two values of “black” and “black”. This error diffusion method uses the gradation value g (i, j) of the multi-valued image.
When binarizing by comparing with a fixed threshold,
The binarization error generated by the binarization is diffused to the gradation value g (i, j) of the neighboring pixel, and the corrected gradation value G (i, j,
j) is sequentially binarized. Here, a binarized symbol (“white” or “black”), a diffusion error de (i,
j) and the corrected gradation value G (i, j) are expressed by the following equations (1) to (1).
(3) required. “White” (G (i, j) <th) s (i, j) = ｛(1) “black” (th ≦ G (i, j)) where th; gs _max / 2) gs _max ; gradation maximum value G (i, j) (G (i, j) <th) de (i, j) = ｛G (i, j) −gs _max (th ≦ G ( i, j)) ··· (2 ) G (i, j) = g (i, j) + a 1 de (i-1, j) + a 2 de (i + 1, j-1) + a 3 de (i, j−1) + a ₄ de (i−1, j−1) (3) where ai (i = 1 to 4); error diffusion coefficient (Σai = 1) As a method of encoding a binarized binarized image (that is, a binary image), an arithmetic encoding method described in Reference 2 is known. FIG. 2 shows a method of generating a binarized image from a multi-valued image by an error diffusion method and encoding the binarized image by an arithmetic coding method. FIG.
1 is a functional block diagram of an encoding device for implementing a conventional image encoding method.

[0003] The coding apparatus includes an error diffusion binarization processing means 10 and an arithmetic coding means 20 comprising a symbol appearance probability estimating means 21 and an arithmetic code forming means 22. In the error diffusion binarization processing means 10, (1) to
Based on the expression (3), binarization of the target image (pixel to be coded) in the input multi-valued image D1 and diffusion of an error from the target pixel to neighboring pixels are performed, and the binarized image D2 is obtained. Then, the binarized image D2 is provided to the symbol appearance probability estimating means 21 and the arithmetic code forming means 22.

The symbol appearance probability estimating means 21 receives the binarized image D2 and uses a "black" or "white" binarized symbol that has been binarized before the target pixel. The symbol appearance probability value is obtained by estimating the appearance probability of the appearing binarized symbol, and the symbol appearance probability value is provided to the arithmetic code forming means 22. Then, the arithmetic code constructing unit 22 encodes the binary image D2 using the symbol appearance probability value as an encoding parameter to generate the code D3 by the arithmetic encoding method.

[0005]

However, in the conventional image encoding method, the error diffusion binarization processing means 10 has a very small binarization error with respect to the binarization threshold.
In some cases, the binarization has little effect on the image quality whether it is a “white” symbol or a “black” symbol. In such a case, if the arithmetic encoding means 20 encodes a predetermined symbol determined by the binarization rule, the code length may be greatly affected. Therefore, in the conventional image encoding method, the binarization processing is not appropriate for efficiently configuring the arithmetic code, and an image encoding method that is technically sufficiently satisfactory is provided. Was difficult.

The present invention provides a multi-valued image coding method which solves the problem of the prior art that the conventional binarizing method is not suitable for data compression using arithmetic codes. Things.

[0007]

According to the present invention, in order to solve the above-mentioned problems, based on an input binary image, "dots present" or "dots absent" appearing at a pixel of interest to be encoded. In the image encoding method for encoding the binary image using the symbol appearance probability value as an encoding parameter, the symbol appearance probability value is obtained by estimating the appearance probability of the binary symbol We have taken such measures. That is, a corrected gradation value is obtained based on the error diffusion binarization processing, the corrected gradation value, the binarization threshold and the symbol appearance probability value are input, and the distance between the corrected gradation value and the binarization threshold is calculated. Is smaller than a predetermined distance evaluation parameter and when it is determined that the code length is longer than the code length defined by the predetermined code length evaluation parameter, the value of the input binary image is corrected and encoded. Like that.

[0008]

According to the present invention, since the image encoding method is configured as described above, a symbol appearance probability value is obtained from an input binary image, and the symbol appearance probability value is used as an encoding parameter to perform arithmetic operation. The binary image is encoded by an encoding method. At this time, when it is determined that the distance between the corrected gradation value and the binarization threshold is smaller than the distance evaluation parameter and the code length is longer, the value of the input binary image is used to determine whether the logic of the symbol is inverted or not. After being corrected by adjusting the diffusion error or the like, it is encoded by an arithmetic encoding method. Thereby, encoding of a binary image can be performed at a high compression rate. Therefore, the above problem can be solved.

[0009]

FIG. 1 shows an embodiment of the present invention and is a functional block diagram of an encoding apparatus for implementing an image encoding method. This encoding device includes error diffusion binarization processing means 3
0 and a binarized symbol / diffusion error correcting means 40,
An arithmetic coding unit 50 is connected to the output side of the error diffusion binarization processing unit 30. The arithmetic coding unit 50 includes a symbol appearance probability estimating unit 51 and an arithmetic code forming unit 52.
The output side of the symbol appearance probability estimating means 51 is connected to the binarized symbol / diffusion error correcting means 40.

The error diffusion binarization processing means 30 binarizes the pixel of interest in the input multi-valued image D11, diffuses a binarization error from the pixel of interest to neighboring pixels, and outputs the binarized image ( That is, it has a function of generating a binary image) D12. The binarized symbol / diffusion error correction unit 40 calculates the symbol appearance probabilities P _W (white) and P _B (black) estimated by the symbol appearance probability estimation unit 51 and the corrected gradation value G (i,
j) and the function of correcting the binarized symbol and the diffusion error in the error diffusion binarization processing means 30 according to the binarization threshold th.

The arithmetic coding means 50 has a function of inputting the binary image D12 from the error diffusion binary processing means 30 and forming an arithmetic code. This arithmetic coding means 50
Is a function of inputting the binarized image D12 and estimating the symbol appearance probabilities P _W (white) and P _B (black) using, for example, a reference image as shown in FIG. have. The arithmetic code constructing means 52
Has a function of using the symbol appearance probabilities P _W (white) and P _B (black) estimated by the symbol appearance probability estimating means 51 as coding parameters to form the code D13 using an arithmetic coding method. . FIG. 4 is a flowchart showing a processing procedure of the encoding device shown in FIG. 1, and the image encoding method of the present embodiment will be described with reference to FIG. FIG.
S1 to S19 in the drawing indicate each processing step.

First, in steps S1 and S2, the arithmetic coding means 50 is initialized, and the position (i, j) of the pixel of interest in the input multi-valued image D11 is initialized. In step S3, the symbol appearance probability estimating means 51 estimates the symbol appearance probabilities P _B and P _W at the pixel position (i, j) from the binarized symbols that have already appeared. Then, in step S4, the error diffusion binarization processing means 30 calculates the corrected gradation value G based on the equation (3).
(I, j) is obtained.

In steps S5 to S7, error diffusion 2
The binarization unit 30 binarizes a modified gradation value G (i, j) a binarization threshold value _{th (= gs max / 2)} . That is, in step S5, modified gray scale value G (i, j) is the binary threshold value th (= gs _max / 2) is determined is greater than or not, "the k in step S6 when large B (black) And when smaller, k is set to "W (white)" in step S7. Then, in steps S8 and S9, the error diffusion binarization processing means 30 calculates the error diffusion de (i, j) based on the equation (2).

In steps S10 to S13, the binarized symbol / diffusion error correcting means 40 sets the corrected gradation value G
Distance between (i, j) and binarization threshold th | G (i, j) -t
When it is determined that h | is smaller than the distance evaluation parameter η and the code length is longer (P _k <ε, ε; code length evaluation parameter), it is determined whether or not k = “B” in step S11. Steps S12 and S
13, the binary symbol “W (white)” or “B
(Black) "and the diffusion error de (i, j)
To correct.

In steps S14 and S15, if the condition of step S10 is satisfied, the arithmetic code forming means 52 forms an arithmetic code with the appearance probability 1- _Pk corresponding to the inverted symbol in step S14. On the other hand, if the condition of step S10 is not satisfied, an arithmetic code is formed by the appearance probability P _k in step S15. And step S
In steps S16 to S19, the coordinate value (i, j) of the pixel of interest is controlled. That is, in step S16, it is determined whether or not i of the coordinate values (i, j) is smaller than _xmax . If smaller, +1 is added to i in step S17 and step S17 is performed.
Return to 3. When i is larger than x _{max in} step S16, it is determined whether or not j <y _{max in} step S18.
When j is less than y _max, as well as a i is set to 0, the process returns to step S3 and adds +1 to j, in step S18, j is the larger than y _max, and ends the encoding process.

In the decoding, for example, as shown in FIG. 3, if the reference image for estimating the symbol appearance probability is the same as that of the coding, the inverse operation of the coding is performed to obtain the original binary image. Can be obtained. As described above, in this embodiment, when the binarization error with respect to the binarization threshold th is small and the symbol appearing in the binarization has a large effect on the code length, the binarization symbol / diffusion error correction is performed. The symbol and the diffusion error appearing in the binarization are controlled by the means 40, and a binarization process suitable for data compression using an arithmetic code is performed based on the control result. Therefore, without deteriorating the image quality of the binarized image D12,
2 can be encoded at a high compression rate, thereby improving the data compression performance of the binarized image D12.

Next, FIGS. 5A and 5B show the results of evaluation of the present embodiment and the conventional method by simulation. FIG. 5A shows a binarized image obtained by a conventional method.
(B) shows a binarized image according to the present embodiment. They are,
Evaluation was made based on the following simulation conditions. [Simulation conditions] Multi-valued image: 256 gradations (8 bits), 512 × 512 Reference pixels for symbol appearance probability estimation: around 10 pixels (see FIG. 3) ε = 0.07, η = 16 FIG. As shown in (a) and (b), the code amount of the present embodiment is reduced by about 13% from the code amount of the conventional method, although there is almost no difference in image quality between the conventional method and the present embodiment. I was able to. Note that the present invention is not limited to the above-described embodiment. For example, the processing procedure of FIG. 4 is changed to another procedure.
Various modifications are possible.

[0018]

As described above in detail, according to the present invention, when it is determined that the distance between the corrected gradation value and the binarization threshold is smaller than the distance evaluation parameter and the code length becomes longer, the symbol The value of the input binary image is corrected by inversion of logic, adjustment of diffusion error, and the like, and then encoded, so that the input binary image can be encoded without deteriorating the image quality of the input binary image. Can be encoded with a high compression rate. Therefore, the data compression performance of the input binary image can be improved.

[Brief description of the drawings]

FIG. 1 is a functional block diagram showing an encoding device for implementing an image encoding method according to an embodiment of the present invention.

FIG. 2 is a functional block diagram showing an encoding device for implementing a conventional image encoding method.

FIG. 3 is a diagram showing reference pixels for estimating a symbol appearance probability in the encoding device of FIG. 1;

FIG. 4 is a flowchart illustrating a processing procedure of an image encoding method according to the present embodiment using the encoding device of FIG. 1;

FIG. 5 is a diagram illustrating simulation results of the present embodiment and a conventional image encoding method.

[Explanation of symbols]

 Reference Signs List 30 error diffusion binarization processing means 40 binary symbol / diffusion error correction means 50 arithmetic coding means 51 symbol appearance probability estimating means 52 arithmetic code construction means D11 multi-level image D12 binary image D13 code

Claims (4)

(57) [Claims] A symbol appearance probability value is estimated based on an input binary image by estimating an appearance probability of a binary symbol “dot present” or “dot absent” appearing at a pixel of interest to be encoded. In an image encoding method for encoding the binary image using the symbol appearance probability value as an encoding parameter by an arithmetic encoding method, a corrected gradation value is obtained based on error diffusion binarization processing. A corrected gradation value, a binarization threshold and the symbol appearance probability value are input, and a distance between the corrected gradation value and the binarization threshold is smaller than a predetermined distance evaluation parameter, and a code length is a predetermined code length evaluation parameter. An image encoding method characterized in that, when it is determined that the length becomes longer than the code length defined by the following, the value of the input binary image is corrected and encoded. 2. The image encoding method according to claim 1, wherein in correcting the value of the binary image, the logic of a symbol of the binary image is inverted. 3. The image encoding method according to claim 1, wherein the binary image is binarized by the binarization threshold using an error diffusion method, and a pixel from the pixel of interest is shifted to a neighboring pixel. Is generated by performing error diffusion to the binary image, and inverts the logic of the symbol of the binary image to obtain the value of the binary image.
And the neighborhood corresponding to the inverted symbol
An image encoding method , comprising: adjusting the diffusion error for pixels . 4. The image encoding method according to claim 1, wherein the binary image is binarized by the binarization threshold using an error diffusion method, and a pixel from the pixel of interest is shifted to a neighboring pixel. By performing error diffusion on the binary image, inverting the logic of the symbol of the binary image, adjusting the diffusion error to a value in a numerical range that the diffusion error can take in accordance with the inverted symbol, An image coding method comprising correcting an image value.