JPH04290177A - Binary coding processing method for gradational image information - Google Patents

Abstract

PURPOSE:To improve the deterioration of contour information in the binary coding process of gradational image information. CONSTITUTION:The method consists of a means which converts the inputted gradational image information into binary data by using a single threshold value, a contour information extraction processing means which detects variation in the gradational image information, decides and extracts a part whose variation quantity is larger than a previously specified value as the contour part of the image, and converts it into binary data, and a composing means which composes information of the binary-coded image information and binary-coded contour information.

Description

[Detailed description of the invention]

0001

[Field of Industrial Application] The present invention relates to a method for binarizing multi-tone image information read by a device having an image input function such as an image reader or a facsimile machine, and in particular to a method for converting contours in an image by binarizing the information. The present invention relates to a binarization processing method for improving information deterioration.

0002

[Prior art] With the shift to multimedia in OA equipment, etc.
The importance of storing and transmitting image information has been increasing in recent years. The storage capacity and transmission capacity of current systems that handle image information are generally not sufficient compared to the amount of image information, and image information often has to be handled in a compressed manner.

[0003] There are various methods for compressing image information depending on the purpose and the nature of the image information, and one of them is to express the gradation of the original image with two gradations (see Figure 4). A specific example is shown). Representing an image with two gradations is
This method not only compresses the amount of information, but also has the advantage that it can be easily output using an image output device that can handle only two gradations, such as a dot printer or a liquid crystal display device.

[0004] Binarization of images is appropriate or inappropriate depending on the type of image. Images such as characters and drawings can originally be expressed in binary values, and there are often no problems even if they are expressed in binary values. However, human images generally contain a lot of important information in a wide range of levels with multiple gradations, and expressing this information in binary terms usually results in an unnatural image.

[0005] In particular, when considering facial expression, it is necessary to determine a binarization threshold that is appropriate for representing facial expressions within the range of changes in gradation values within the face, and a threshold value that is appropriate for expressing facial expressions within the range of changes in gradation values between the face and background. Normally, the threshold value is different from the one appropriate for representing the outline of the face. For example, when a facial expression is properly expressed, the overall image becomes dark and the outline becomes unclear, and when an attempt is made to express the outline clearly, the expression tends to become too bright.

[0006] In this way, when the absolute value of the gradation is binarized using a single threshold value, it is difficult to express both important information for the human image, such as the expression of facial expression and the expression of the outline of the face. Can not do it.

[0007]

SUMMARY OF THE INVENTION In the conventional binarization processing of multi-tone images, important contour information is sometimes ignored. An object of the present invention is to allow important contour information to be incorporated into binarized information without being ignored.

[0008]

[Means for Solving the Problems] In addition to the conventional binarization processing means for multi-gradation image information using a single threshold value, the present invention also provides a means for binarizing multi-gradation image information using a single threshold value. The area where the amount of change exceeds a pre-specified constant value is determined to be an outline, is extracted and binarized, and is combined with the result of binarization by the binarization processing means using a threshold value. By doing so, binary image information is obtained in which important contours are emphasized.

FIG. 1 shows the basic configuration of the present invention. Figure 1
1, an image input device reads an image using an image sensor such as a CCD and outputs it as multi-gradation image information in a digital signal format.

Reference numeral 2 denotes an image information file in which multivalued image information is stored and read out as necessary. 3 is an image processing device that performs image processing, display, printing, storage, transmission, etc. 4 is a binarization processing means, which binarizes input multi-gradation image information using a single set threshold value and outputs the binarization processing means.

Reference numeral 5 denotes a contour information extraction processing means, which compares the tone values of each two adjacent pixels in the input multi-tone image information, and calculates the difference between the tone values of the two mismatched pixels. It is determined whether the amount of change is greater than a pre-specified threshold, and if it is, it is determined that it is a contour part, and the tone values of the two pixels of the pair are binarized based on their relative magnitude. Output.

[0012] 6 is a synthesis means, and a binarization processing means 4
and each output of the contour information extraction processing means 5 are synthesized pixel-wise. The synthesis method is such that, for example, if both output values are equal, that value is output, and if the output values are different, the output value of the contour information extraction processing means is selected and output.

[0013]

[Operation] The operation of the present invention will be explained using a specific example shown in FIG. Figure 2
(a) is input multi-gradation image information, and shows an example of the gradation values of each pixel arranged in a certain scanning line.

The binarization processing means 4 in FIG. 1 binarizes the multi-gradation image information in FIG. 2(a) using a threshold value ES0, and
If it is larger than 0, it is output as "1", and if it is smaller, it is output as "0". The contour information extraction processing means 5 in FIG.
), (Pm, Pm+1), (Pn, Pn+1)
Detect.

[0015] Figure 2 (b) is a binary version of this, in which the value of the pixel that is relatively closer to white is converted to "1", and the value of the pixel that is closer to black is converted to "0". do. (c) in Figure 2 is
The output of the binarization processing means 4, (d) in FIG. 2 shows the output of the contour information extraction processing means, and (f) in FIG. 2 shows the binarized image information as a result of synthesis by the synthesis means 6.

[0016] In this way, a black line corresponding to the outline is formed in the white area of the binarized image information, and a white line is formed in the black area, thereby preventing information regarding the outline from disappearing.

[0017]

Embodiment FIG. 3 shows the configuration of one embodiment of the present invention. Although this embodiment is shown as a hardware configuration, it goes without saying that the same logic can be easily implemented in software.

In FIG. 3, 7 is an image memory that stores multi-tone image information. Here, the data width of the multi-gradation image information is 1 byte per pixel, and 256 gradations can be expressed. The pixel address is represented by Xi in the horizontal direction and Yj in the vertical direction. Furthermore, the pixel at address Xi, Yj is represented by Pi, j, and its pixel value is Di,
j. And the adjacent pixels in the X direction are Pi+1,j
, the pixel value becomes Di+1,j.

8 is an address generation section. 9 is Xi
This is an X address counter that generates an address. 10
is the Y address counter that generates the Yi address. 11 is a binarization processing section.

12 is a comparator that binarizes Dij with a threshold value ES0, and when Dij≧ES0 as a binary output, B
It outputs ij=1, and when Dij<ES0, it outputs Bij=0. 13 is a contour extraction processing section. 14 is a Dij register in which a pixel value Dij is set.

15 is a Di+1,j register in which the pixel value Di+1,j is set. 16 is Dij and D
This is a subtracter that detects the difference (change amount) Dij-Di+1,j between i+1,j. 17 is the subtraction result (Dij-D
i+1,j), and the threshold ES1, Dij,D
Binary output of i+1,j respectively Mij, Mi+1,j
This is a binary logic circuit that converts Here｜Dij-
Di+1,j |≧ES1, and if Dij>Di+1,j, then Mij=1, Mi+
1,j =0Dij<Di+1,j, then Mij
=0, Mi+1, j =1 are output.

[0022] Also, |Dij−Di+1,j |<ES
If it is 1, the Mij, Mi+1,j output is disabled. 18 is a Mij register that stores Mij. 19 is Mi+1,j that stores Mi+1,j
It is a register.

20 is Bij and Mij, Bi+1,j
and Mi+1,j, respectively, and M
ij, Mi+1,j corresponds to Bij, Bi+1,j
Replace. As a result, Bij≠Mij or Bi
+1,j ≠Mi+1,j, then Bij,
Bi+1,j is replaced by Mij,Mi+1,j, and Bij=Mij or Bi+1,j =Mi+
1,j, Bij, Bi+1,j is not changed.

The Xi counter 9 of the address generating section 8 performs address scanning in the X direction by counting clocks, and when the count reaches a full count, the Yi counter 10 is incremented by 1 to advance the address by one in the Y direction. The address Xi of the pixel Pij continuously generated in this way
, Yj are applied to the image memory 7, Pij is scanned, and the pixel value Dij is read out. Dij is binarized by the binarization processing unit 11 to become Bij, and the sudden change part is detected by the contour extraction processing unit 13 and binarized contour information Mij, Mi+1,j is obtained, and the synthesis unit 20 DeB
It is combined with ij, Bi+1, j and output.

The value of the threshold ES1 that determines the extraction level of a sudden change part in multi-tone image information can be set arbitrarily. Furthermore, the extraction of contour information from multivalued image information is not limited to the X direction, but can also be performed two-dimensionally in both the X and Y directions.

[0026]

[Effects of the Invention] According to the present invention, important outline information can be retained when binarizing multi-gradation images such as images that are entirely whitish, images that are entirely black, or photographs that include various gradations. The amount of effective information in the digitized image information is increased, and the usefulness of the image information is significantly improved.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the basic configuration of the present invention.

FIG. 2 is an explanatory diagram showing a specific example of binarization processing for explaining the effect of the present invention.

FIG. 3 is a configuration diagram of an embodiment of the present invention.

FIG. 4 is an explanatory diagram of a specific example of conventional binarization processing.

[Explanation of symbols]

1 Image input device 2 Image information file 3 Image processing device 4 Binarization processing device 5 Contour information extraction processing means 6. Synthesis means

Claims (3)

[Claims] Claim 1: In a method for binarizing multi-gradation image information, input multi-gradation image information is binarized using a single threshold value.
A binarization processing means (4) for converting into a value, detects a change in the input multi-tone image information, and determines a portion where the amount of change is larger than a predetermined value to be an outline of the image. It is equipped with a contour information extraction processing means (5) for extracting and binarizing the image information, and a synthesizing means (6) for synthesizing the binarized image information and the binarized contour information. 1. A method for binarizing multi-gradation image information, characterized in that gradation image information is binarized with edge emphasis. 2. The multi-gradation image according to claim 1, wherein the detection of a change point in the multi-gradation image information is performed one-dimensionally between adjacent pixels along a scanning line of the image. Information binarization processing method. 3. The binarization process of multi-gradation image information according to claim 1, characterized in that the extraction of changing points in the multi-gradation image information is performed two-dimensionally between vertically and horizontally adjacent pixels. Method.