JPH0474283A - Method and device for picture processing - Google Patents

Abstract

PURPOSE:To obtain a binary picture without losing form information of a variable density picture by generating a histogram of the difference between adjacent maximum and minimum values to set a critical value of the histogram and obtain a threshold for binarization in accordance with maximum and minimum values between which the difference is larger than the critical value. CONSTITUTION:Values and positions of extreme values of the variable density picture in a picture memory circuit 3 are detected and stored, and the difference between adjacent extreme values is obtained by a difference detecting circuit 5, and the histogram of the difference is obtained by a histogram generating circuit 6, and a value CT at the bottom point of the histogram is obtained by a bottom point detecting circuit 7. Extreme values related to the image are obtained from the obtained value CT by a comparing and selecting circuit 8, and extreme values are used to determine the threshold by a threshold calculating circuit 9, and the variable density picture is converted to a binary picture in accordance with variable density picture information and the threshold. Thus, the binary picture is obtained without losing picture form information of the variable density picture.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image processing method and apparatus for converting a grayscale image input from an image element into a binary image.

(Prior art) Normally, when image processing is performed, a grayscale image is converted into a binary image and various analyzes are performed. Conventionally, in such binarization, the same threshold is applied to the entire screen. Apply the value and
Binarization processing was performed.

In this way, if the same threshold value is applied to the entire screen, when the grayscale image input from the image sensor is converted into a binary image, the shape information of the original grayscale image will be lost. Further analysis may not be possible. This occurs because a grayscale image has grayscale information of 256 gradations when captured by a normal image sensor, whereas a binary image has grayscale information of 2 grayscales. In other words, when converting 256-level gradation information into 2-gradation gradation information, information is lost, and the shape information of the figure is also lost.

In order to prevent such information loss, conventionally it has been very important to perform processing to enhance the contrast of the image before binarization processing, and to equalize the illumination for inputting the image to the imaging element. It was a great technique. However, even if such careful attention is taken to input an image to an image sensor and contrast enhancement processing is performed, in reality, there are many cases where the binarization is not successful. This is currently an important issue in application development.

An example in which conventional binarization processing is applied will be explained using FIG. 4.

In FIG. 4, (a) is a grayscale image input from the image sensor, (b) is a one-dimensional density histogram at X-X, and (C), (d), and (e) are threshold values, respectively. is A.

Binary images for B and C are shown.

As can be seen from the example in Figure 4, in conventional binarization processing, the 'a gradation image shape information is lost, and the binary image becomes completely different from the original image. . It goes without saying that if such a loss of shape information occurs, correct results cannot be obtained in image processing.As a way to solve the above problem, the image input from the image sensor etc. is For each scanning line or each vertical scanning line, find the maximum and minimum values of the density value, calculate the binarization threshold from the adjacent maximum and minimum values, and set the threshold value to the corresponding maximum and minimum values. The threshold value between the local maximum and local minimum values is 2(l
[The zero method, which has already been proposed by the present inventor as a method for obtaining images, is a very superior method compared to conventional methods, but for images with a lot of noise, as preprocessing,
If noise removal is not performed carefully, there are drawbacks in that noise may be added to the binary image, and shape information of the image is likely to be lost in the case of fine line images.

(Problems to be Solved by the Invention) In view of the above-mentioned problems, the present invention provides a binary image detection method that does not lose the shape information of the grayscale image input from the imaging element and is resistant to noise. or for the purpose of providing such equipment.

(Means for Solving the Problems) The present invention is a method of processing an image input from an image sensor etc. into a binarized image, in which a manually generated image is processed for each horizontal scanning line and each vertical scanning line. Find the local maximum value and local minimum value of the density value, calculate the difference between the adjacent local maximum value and local minimum value, create a histogram of this difference (Δa), and set the critical value (trough point: C1) of the histogram. Then, find the local maximum value and local minimum value where Δa is larger than Ca, calculate the threshold value for binarization from this maximum value and local minimum value, and create an image that takes the local maximum value and local minimum value that are adjacent to the threshold value. This is an image processing method characterized by at least the step of obtaining a binarized image as a threshold value of the input signal, and an apparatus for processing an image manually inputted from an image sensor or the like into a binarized image. Calculate the maximum and minimum density values for each horizontal scanning line and each vertical scanning line of the image.
An image characterized by having at least a critical value setting circuit that creates a histogram of the absolute value (Δa) of the difference between adjacent local maximum values and local minimum values and sets a critical value (trough point: Cy) of the histogram. It is a processing device.

The present invention will be explained using FIG. 2. Figure 2 shows a one-dimensional density histogram in a horizontal (or vertical) scanning line of an image obtained by an image sensor.
Xl+Xt+..., χ3. X. , . . . indicate pixels whose density value takes an extreme value, and 1. χ2゜Xi+Xi
The density value at +1... is al+82...ai+a
i. 1. ...and...

When an image is captured by an image sensor and a one-dimensional density histogram is created, as shown in FIG. 2, there are cases where the difference (hereinafter referred to as contrast) between the maximum value and minimum value adjacent to each other is small and the difference is large. A small contrast represents noise, a large contrast represents an image. As a method to find the bifurcation point between noise and image contrast,
The present invention proposes a method using a contrast histogram.

Usually, noise and image contrast can be clearly distinguished.

Therefore, the contrast histogram has bimodal properties as shown in FIG.

In FIG. 3, a value smaller than the valley point value Cr indicates noise, and a value greater than the valley point value Cr indicates image contrast.

Now, as a method for finding the branch point C7 between noise and image contrast from such a contrast histogram, we propose the following three methods depending on how samples are selected.

(1) A method of determining each horizontal scanning line or vertical scanning line. In this case, the value of C1 changes for each scanning line.

(2) A method of determining in the entire horizontal scanning direction or vertical scanning direction. In this case, the value of C0 is fixed in each scanning direction.

(3) A method of calculating in the entire horizontal scanning direction and vertical scanning direction. In this case, the value of C7 takes a single value in the horizontal and vertical scanning directions.

Each of the above methods has advantages and disadvantages, but if you want to shorten the processing time, method (3) is better.

Next, a method for determining the threshold value T II L will be described.

In Fig. 2, the extreme value that satisfies formula (1) is noise, formula (2), ,, -a, l > cTi= 1.2・------
Extreme values that satisfy (2) represent images.

Now, when we extract only the extreme values that satisfy equation (2) from the extreme values of a++az+..., it becomes a+'+az'
r...and al' +8! ' l...
Let the pixels corresponding to the density values of . . . be ×1, l .

During this period, between the adjacent extreme values Xi'~xt-+(i・1
．． 2...) is determined by the following equation.

For example, if equation (3) is applied to FIG. 2 with 0.5, the result will be as shown in FIG. 1.

In FIG. 1, the threshold value for the area considered to be noise has not been determined. The threshold value for such an area is determined from the values of the previous and subsequent threshold values. For example, the threshold value in area A in FIG. 1 is the threshold value THL' before and after it.
, THL, by one of the following methods.

1) Set the threshold to THL' or THL''2)
The threshold value is the average value of THL' and THL" Which of the above methods is better depends on the image, but
The experimental results show that there is not much difference.

When the present invention is applied to the image in FIG. 4 with 0.5, the result is as shown in FIG. 5.

From Figures 1 and 5, the threshold value changes depending on the density value around the image; the higher the density, the higher the threshold value, and the lower the density, the lower the threshold value (the situation is different). I understand.

In this way, the threshold value changes depending on the density value around the image. This means that humans are doing something similar to determining the boundaries of an image, so if the present invention is used, the image shape information of the gray scale image will be lost compared to the conventional method. It can be seen that a binary image can be obtained without any movement.

In this way, the threshold value THL is determined for each horizontal or vertical scanning line, and the threshold value THL is determined for each horizontal or vertical scanning line.
1.2. ), the determined threshold value T 1
A pixel with a density of 1 L or more is treated as white or black, and a pixel with a density of less than T 11 L is treated as black (or white). As a result, one binary image is obtained for each horizontal or vertical scanning line.

In addition to the method of using the binary images obtained as described above as they are, the present invention proposes a method of calculating the AND or OR of the binary images as described below. By doing so, generation of a clean binary image from which noise has been removed and generation of a binary image from a blurred image can be achieved at a higher level. However, on the other hand, it also has the disadvantage that the processing time becomes longer.

Therefore, whether or not to apply the following method will be determined depending on the purpose of image processing.

Hereinafter, a method of obtaining a binary image by performing a logical product and a logical sum of two images will be described.

Now, a binary image is obtained by performing a logical product (AND) or a logical sum (OR) between corresponding pixels of the two binary images obtained as described above. Now, assuming that the background is O (white) and the image is 1 (black), a binary image can be obtained by performing the AND and OR of the two as follows.

Cij is a logical product, C'1 is a binary image obtained by a logical sum, a, , is a binary image obtained by horizontal scanning, bi
J is a binary image obtained by vertical scanning, Cij+C'
i□ indicates a 2(! image) obtained by logical product and logical sum of these binary images, and N and M indicate the number of pixels in the horizontal and vertical directions, respectively.

The effect of taking logical products and logical sums as above is explained in the sixth section.
This will be explained using figures.

In FIG. 6, (a) is a grayscale image input from the image sensor, (b) is an image 2 (a) obtained by horizontal scanning, and (C) is a binary image obtained by vertical scanning, ( d)
represents the binary image obtained by the AND of images (b) and (C), and (e) represents the binary image obtained by the logical sum of the images (b) and (C). There is.

In Figure 6, ■ and ■ are line images with contrast close to the noise level, ■'1■' is the binary image, and ■ and ■ are the line images that have a contrast close to the noise level, and ■ and ■ are the line images that have a contrast close to the noise level. This shows the noise generated.

A binary image (b) obtained by scanning in the horizontal direction usually has high sensitivity to the horizontal direction, and therefore tends to have horizontal noise as shown by ■. Furthermore, the sensitivity in the vertical direction is conversely lower than the sensitivity in the horizontal direction, and line images with contrast close to the noise level are difficult to detect.

In the case of a binary image (C) obtained by scanning in the vertical direction, the sensitivity in the vertical direction is high and the sensitivity in the horizontal direction is low, contrary to the case of scanning in the horizontal direction.

Therefore, a binary image as shown in (C) is obtained.

Therefore, the image C4j obtained by taking the logical product of the binary image aij and t)iJ is a line image with a contrast close to the noise level, instead of being a clean image with the noises ■ and ■ removed. Information is lost. When the binary images aij and bij are logically summed, the obtained image c
′, the information of the line image with a contrast close to the noise level is not lost, but the noises ■ and ■ remain without being removed.

In this way, by taking the logical product, the information of the line image with poor contrast is lost, but a clean binary image with noise removed is obtained. Although a binary image is obtained without any loss of noise, noise remains without being removed.

In this way, the 2 obtained by taking logical products and logical sums
Each value image has its advantages and disadvantages, but depending on the case,
By using them properly, extremely effective processing can be achieved.

(Example) (Example-1) An embodiment of the present invention will be described using FIG. 7.

The grayscale image input from the CC camera (1) is A/D
The conversion circuit (2) converts the grayscale values from analog values to digital values, and the grayscale image of the 0 image memory circuit (3) is stored in the image memory circuit (3). Each scanning line is taken out, and the value and position of the extreme value are detected and stored in the extreme value detection circuit (5). From the stored extreme values, a difference detection circuit (5) calculates the difference between adjacent extreme values, and a histogram creation circuit (6)
The histogram of the difference is obtained by using the trough detection circuit (
7), the valley point (l[Cr) of the histogram is determined.

From the obtained value of Cr, the computation of equation (2) is performed by the comparison and selection circuit (8), and the value of the extreme value related to the image is obtained. The threshold value calculation circuit (9) uses the obtained extreme value to determine the threshold value from equation (3).

In the black and white determination circuit 00), the grayscale image is converted into a binary image based on the grayscale image information from the image memory circuit (3) and the threshold value from the threshold calculation circuit (9). The converted binary image is stored in the binary image memory circuit (11) and output from the output circuit 0 as required.

In addition, as a method of obtaining the valley point C1 from the histogram creation circuit, there is a method of obtaining C7 for each horizontal or vertical scanning line, and a method of obtaining C7 for each horizontal or vertical scanning line.
As mentioned in the detailed description of the invention, there are methods for finding C7 and finding one C1 by scanning all the horizontal and vertical scanning lines, but the structure is the same as that shown in FIG. Since this method can be implemented using various methods, the explanation will be omitted.

(Example 2) In Example 1, a method for obtaining a binary image by scanning the image in the horizontal or vertical direction was described. A method of obtaining a binary image by logical product or logical sum between images will be explained using FIG.

In Figure 8, (1) to (11'), (11'
) is the same as that in FIG. 7, so the explanation will be omitted.

On the logical product and logical sum operation circuit side, formula (4) or ( The calculation of equation 5) is performed, and the 2 (+!! image) is obtained, stored in the 2 (II image storage circuit 04), and output from the output circuit surface as required.

(Effect of the invention) FIG. 9 shows an actual application example of Example-2.

FIG. 9(a) shows an fA light image captured from a CCO camera, FIG. 9(C) shows a binary image obtained by the conventional binarization method, and FIG. A binary image obtained by a binarization method (a method using logical product) is shown.

From FIG. 9, it can be seen that in the conventional method, the image shape information of the grayscale image is considerably lost, whereas according to the present invention, a binary image is obtained without losing much of it. I understand that.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing a threshold value when one example of the present invention is applied to an example of a grayscale image, FIG. 2 is a diagram showing an example of a grayscale image, and FIG.
The figure is a histogram showing an example of the absolute value of the difference between the maximum value and the minimum value that are adjacent to each other. The figure shows an outline of a grayscale image and a binarized image when one example of the method of the present invention is applied, Fig. 6 shows an outline of an example of applying logical product and logical sum operations, and Fig. 7 shows an example of the implementation. FIG. 8 is a diagram schematically showing the arrangement of each circuit in Example-1, FIG. 9 is a diagram schematically showing the arrangement of each circuit in Example-2, and FIG. 9 is a schematic diagram showing the application result in Example-2. This is a diagram. (None) Ichimatsu〆O: Shishiba・・柃＠、儂朂（C） Ichimoku・・－１１１１１一一く 9 Engraving of a mime drawing (no changes to the content) Illustration 15 (Subushi Inuguchi 1゜2゜3゜Procedural amendment (method) % formula % Display of the case 1990 Patent m)!l 8 s No. 664 Name of the invention Image processing method and its device Person who corrects the device Relationship with the case Patent applicant Date No. 6, 2-2 Dojimahama, Kita-ku, Osaka, Contents of amendment (1) Engravings of Figures 4, 5, 6, and 9 of the drawings originally attached to the IjJl book, and the attached sheet. As per (no change in content)

Claims (2)

[Claims] (1) A method of processing an image input from an image sensor etc. into a binarized image, which converts the input image into maximum and minimum density values for each horizontal scanning line and each vertical scanning line. seek,
Find the difference between the local maximum and minimum values that are next to each other, and calculate this difference (Δ
Create the histogram of a), set the critical value (valley point: Cr) of the histogram, find the local maximum value and local minimum value where Δa is larger than Cr, and use the local maximum value and local minimum value to determine the threshold for binarization. An image processing method comprising at least the steps of determining a value, and determining a binarized image using the threshold value as a threshold value of an image that takes adjacent maximum and minimum values. (2) A device that processes an image input from an image sensor etc. into a binarized image, which converts the input image into each horizontal scanning line,
The maximum and minimum values of the density values are determined for each vertical scanning line, and a histogram of the absolute value (Δa) of the difference between the adjacent maximum and minimum values is created, and the critical value of the histogram (trough point: C
An image processing device comprising at least a threshold value setting circuit for setting r).