JP2976796B2 - A method of determining the threshold of an area by image analysis - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of determining a threshold value of a region by image analysis for determining a threshold value of a density range corresponding to each region of an image representing a plurality of regions.

[0002]

2. Description of the Related Art Analyzes are performed to binarize an image of a cross section of a synthetic substance having a plurality of tissues for tissue discrimination, extract a binary image relating to a desired tissue, and obtain the area thereof.
JP-A-60-1000032 and JP-A-2-23
Japanese Patent No. 2550 discloses that when an image is a binary image or a region of several levels of density for tissue classification, a histogram of a density distribution area is created for the image according to a preset density level, and a reference image is obtained. When the examiner inputs a threshold value for determining the density range for each target tissue with respect to this histogram, a histogram is created for a similar image similar to the reference image, and the histogram of the similar image is calculated according to the threshold value of the reference image. A method for automatically determining a threshold is disclosed.

[0003]

However, the above-mentioned threshold value is set at the bottom position of the histogram (the position where the minimum value appears). In the case of a histogram where a clear bottom does not appear, the threshold value is automatically set. It was difficult to determine.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and is intended to match the gray scale of a reference image with the gray scale of a similar image similar to the reference image, or to match the characteristic position of the histogram of the reference image. Based on the difference between the feature positions of similar images,
An object of the present invention is to determine a threshold value of a similar image based on a threshold value of a reference image even in a histogram having no clear bottom. In addition, the feature position of the histogram of the reference image and the feature position of the histogram of the similar image are matched by adjusting the illumination so that the threshold value of the similar image can be determined from the threshold value of the reference image using hardware. Aim.

[0005]

[0006]

Means for Solving the Problems To achieve the above object,
For example, a histogram of a grayscale distribution area is created from images representing a plurality of regions in accordance with a predetermined grayscale level, and a threshold value of a region by image analysis that defines a threshold value representing a density range corresponding to each target region with respect to the histogram. In the decision method,
Determining a feature position representing a feature of the histogram of the reference image;
A threshold value representing a density range corresponding to each target area is determined in the histogram, and a characteristic position of the histogram is obtained for a similar image similar to the reference image. The characteristic position of the histogram of the reference image and the characteristic position of the histogram of the similar image are calculated. The position where the difference is added to the threshold value of the histogram of the reference image is set as the threshold value of the histogram of the similar image.

[0007] A histogram of the distribution area of the gray scale is created in accordance with a predetermined gray scale from images representing a plurality of regions captured by the television camera by illuminating the sample, and the histogram corresponds to each target region. In a method of determining a threshold value of an area by image analysis that determines a threshold value representing a density range to be obtained, a histogram of a reference image and a characteristic position thereof are obtained,
A threshold value representing a density range corresponding to each target region is determined in the histogram, and a histogram and a characteristic position of the similar image similar to the reference image are obtained. The characteristic position of the histogram of the similar image matches the characteristic position of the histogram of the reference image. Lighting is adjusted so that the threshold of the histogram of the reference image is set as the histogram threshold of the similar image.

The feature position is defined as a density value of the area centroid of the histogram.

The characteristic position is a median between an upper limit value and a lower limit value of the density of the histogram.

Further, the characteristic position is set as a lower limit value of the density of the histogram.

[0011]

The histogram of the gray scale distribution area of the reference image and the histogram of the similar image similar to the reference image have similar shapes because the images are similar, but the average is due to the difference in the brightness of the illumination for irradiating the sample. Concentrations are often different. For this reason, if the characteristic position representing the characteristic of the histogram is obtained and the characteristic position of the histogram of the similar image is matched with the characteristic position of the histogram of the reference image, the shapes of both histograms overlap quite well and the threshold value is also at the same position. . Therefore, the threshold determined by the histogram of the reference image
The threshold value of the histogram of the image similar to the reference image can be determined by adding the difference between the feature positions of both histograms. In addition, as a method of superimposing the histogram of the reference image and the histogram of the image similar to the reference image, by adjusting the illumination of the sample from which the similar image is obtained, the feature positions of both histograms can be matched and superimposed.
After matching the feature positions, the threshold of the histogram of the reference image can be used as the threshold of the histogram of the similar image. As the characteristic position, the density value of the area centroid of the histogram, the median of the lower and upper limits of the density of the histogram,
The lower limit of the density of the histogram is used because it represents a feature of the histogram. The threshold value of the histogram thus determined is a threshold value representing a density range corresponding to the area of the reference image and the similar image.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an apparatus for realizing this embodiment. The microscope 1 is provided with an imaging lens attached to an eyepiece and an imaging device 20 for taking an image through the imaging lens. The stage 21 on which the measurement sample is placed moves in the Y and X directions to set a measurement position area of the measurement sample. The control is performed by an auto stage device 15 described later. The auto stage device is disclosed in Japanese Patent Publication No. 1-28992 and Japanese Patent Publication No. 5-70 by the present applicant.
No. 126 provides a detailed description. Also, a moving mechanism 22 for moving up and down by a pulse motor provided on the stand
, The position signal is transmitted to the autofocus device 16 described later, and the movement control signal is received to move the stage 21 in the vertical direction. The illumination device 23 illuminates the measurement sample, and adjusts the intensity of the illumination by the automatic illumination adjustment device 17 described later.

The A / D converter 2 converts input data from the imaging device 20 from analog to digital, and the input buffer 3 temporarily stores the digital data. The bus 4 transmits signals, the program memory 5 stores a program that defines the operation of the apparatus, and the CPU 6 controls the entire apparatus according to the program.

An image processor 7 performs density processing, binarization processing, image analysis, and the like on the input image data. A density image memory 8 stores density image data, and a binarization memory 9 stores binary image data. Store. The output buffer 10 temporarily stores the data to be output, and the D / A converter 11 converts the output data from digital to analog.
2 displays this output data on the screen. A control management computer 14 is connected to the bus 4 via an interface 13, and controls the autostage device 15, the autofocus device 16, the automatic lighting adjustment device 17, the printer 18, and the monitor 19. The auto stage device 15 moves the stage 21 in the X and Y directions so that the measurement position and region of the measurement sample determined by the control management computer 14 are at predetermined measurement points in the field of view of the imaging device 20. The autofocus device 16 controls the moving mechanism 22 so as to obtain a clear image on the imaging device 20, moves the stage 21 in the vertical direction, and automatically focuses. Automatic lighting adjustment device 1
Reference numeral 7 denotes a lighting device 2 based on image analysis of the image processor 7.
The image density is adjusted by controlling the illumination intensity of No. 3.

FIG. 2 shows a grayscale image (multi-valued image) of a metal surface tissue or a living tissue. The grayscale image is a representation of image data input from the imaging device 20 at a predetermined grayscale level (gray level), for example, 256 levels. The grayscale image is processed by the image processor 7, stored in the grayscale image memory 8, and then stored in the CRT 12. Is displayed. In order to inspect the tissue represented by this grayscale image, there is a method of extracting each tissue or a desired tissue and calculating its area.In this case, a threshold is set to classify each tissue, and this threshold is used. Binarization is performed to extract a binary image of a desired tissue, and the area is calculated.

When measuring the composition of a metal surface tissue or a living tissue, a large number of samples are often taken from the same sample or a sample of the same series, and many similar tissues are often examined. The inspector sets the threshold while looking at the screen, and thereafter displays the difference between the reference image and the target image with appropriate parameters, and uses the parameters to set the threshold of the target image based on the threshold set in the reference image. Set automatically.

FIG. 3 is a histogram showing the density distribution of the gray image, where the horizontal axis represents the density level and the vertical axis represents the frequency. The frequency is represented by the ratio of the area having the corresponding density to the entire area of the screen cut out as the inspection target. The grayscale image shown in FIG. 2 is composed of four regions a, b, c, and d, where a represents the darkest and d represents the brightest. FIG. 3 is a histogram of the grayscale image of FIG. 2, in which the area a is the smallest and the area d is the largest. The histogram curve is obtained from the grayscale image shown in FIG. 2 according to a program stored in the program memory 5 in advance.
It is created by the image processor 7 and displayed on the screen of the CRT 12 by the CPU 6.

Using the grayscale image of FIG. 2 as a reference image, the inspector sets a tissue boundary (that is, a threshold) while looking at the histograms of FIGS. 2 and 3. In this case, three thresholds T1, T2, and T3 are set because the region is composed of four regions. A method for automatically setting the threshold value of the histogram of an image similar to the reference image by the image processor 7 based on the threshold values T1 to T3 of the histogram of the reference image will be described below. Note that since the image is binarized by the threshold of the histogram, the threshold of the histogram has the same meaning as the threshold of the image.

Next, the operation until the image of the measurement sample is binarized by the apparatus shown in FIG. 1 will be described with reference to the flowcharts of FIGS. When an intended image is obtained by an optical microscope, first, a measurement target position of a sample is determined (ST1),
Next, a measurement area is extracted from the position (ST2),
Take an image. The positioning of the measurement target and the determination of the area are performed by the auto stage device 15. After the image of the target area is input (ST3), shading correction for correcting unevenness of the image density due to uneven lighting or the like is performed (ST4), the density of each pixel constituting the image is obtained, and the density distribution frequency is represented. A density histogram is created (ST5). Next, a density distribution range is obtained (ST6). Since the density histogram has the horizontal axis as the density as shown in FIG. 3, the density range to be distributed can be easily obtained. The density level of the image is normalized from this density range (ST7). For example, a concentration of 0 to 25
In the case where the histogram is expressed in the range of 5, the density range of the histogram is allocated to the range of 50 to 200 with a margin of an upper limit and a lower limit. Note that a density distribution range is obtained (ST6).
The step of normalizing (ST7) is necessary only for ST12, which will be described later, and is unnecessary when using the threshold determination method of ST11 and ST13. Next, one image is taken out as a reference image, and the inspector determines the threshold value of the reference image by referring to the histogram (ST8). FIG. 3 shows the threshold values thus determined. Next, the characteristic position of the density histogram is obtained (ST9). As the characteristic position, the density value of the center of gravity of the histogram, the density of the median of the upper and lower limits of the histogram, and the density of the lower limit are used.

Next, a method for determining a threshold value of a similar image is selected (ST10). The threshold value of the similar image is determined based on the threshold value of the reference image.There are several methods for determining the threshold value, and when the inspector specifies the method in consideration of the measurement target, the apparatus sets the threshold value of the similar image according to the method. Determined automatically.
In the first method, the threshold value of the similar image is set to the same value as the threshold value of the reference image, and the assignment of the density of the similar image is moved so that the density levels of the characteristic positions of the reference image and the similar image match. This is the method shown in the example (ST11). The second method is to determine the threshold value of the similar image such that the difference between the threshold value and the feature position of the reference image and the similar image is the same, and
This is the method shown in the fourth embodiment (ST12). The third method is a method in which the threshold value of the similar image is set to the same value as the threshold value of the reference image, and the illumination of the sample is adjusted so that the density of the feature position is the same in the reference image and the similar image. The method is (ST13). The reference image and the similar image are binarized based on the threshold value thus determined (ST14).

The first embodiment will be described with reference to FIG. This embodiment sets the threshold value of the similar image to the same value as the threshold value of the reference image,
The density level of the similar image is moved so that the density of the feature position of the reference image and that of the similar image match, and the case where the center of gravity of the histogram is used as the feature position will be described. The threshold value can be determined in the same manner when the median value and the lower limit value of the density are used instead of the center of gravity.

FIG. 6A shows a histogram of a reference image, and the solid line in FIG. 6B shows a histogram of a similar image. Let G1 be the area centroid of the histogram of the reference image, and X1 be its abscissa (density). Also, let the area centroid of the histogram of the similar image be G2 and its abscissa be X2. The difference ΔX = X2−X1 between the two centroids is obtained, and this value is subtracted from the density of each pixel of the similar image. If ΔX is positive, subtraction is performed, and if ΔX is negative, the absolute value of ΔX is added. As a result, the densities of similar images can be shifted by the difference ΔX so that the centers of gravity G1 and G2 of both histograms can be matched. The broken line in (B) indicates a state where the density of each pixel of the similar image is shifted by ΔX to make G1 and G2 coincide.

A second embodiment will be described with reference to FIG. In the present embodiment, the value of the center of gravity is used as the feature position of the histogram. First, a histogram is created for the reference image, and the inspector sets a threshold value for this histogram. Then, the center of gravity of the histogram is obtained. Are set so that the distance on the abscissa is the same on the histogram of the reference image and the similar image. FIG. 7A shows a histogram of a reference image, and FIG. 7B shows a histogram of a similar image.
Let G1 be the area centroid of the histogram of the reference image, and X1 be its abscissa (density). Also, let G2 be the area centroid of the histogram of the similar image, and let X2 be its abscissa.
The difference ΔX = X2−X1 between the two centroids is obtained, and this is added to the abscissa of each of the thresholds T1 to T3 set in the histogram of the reference image.
1 to T3 can be set. The above operation is performed by CPU
6, which is performed by the image processor 7 or the like.

Next, a third embodiment will be described with reference to FIG.
In this embodiment, the characteristic position of the histogram is represented as
The median between the lower limit (darker) and the upper limit (brighter) of the histogram is obtained, and the threshold value of the similar image is set such that the distance between the median value and the threshold value on the abscissa is the same on the histograms of the reference image and the similar image. Set. FIG. 8A shows a histogram of a reference image, and FIG. 8B shows a histogram of a similar image. The median (X1) is calculated from the upper limit and the upper limit of the histogram of the reference image, and the median (X2) of the histogram of the similar image obtained in the same manner is calculated.
By adding X2-X1 to the abscissa of the threshold values T1 to T3 set in the histogram of the reference image, the threshold values T1 to T3 can be set on the histogram of the similar image. This setting is also performed by the CPU 6, the image processor 7, and the like.

Next, a fourth embodiment will be described with reference to FIG.
In this embodiment, the lower limit (darker) of the histogram is obtained as representing the characteristic position of the histogram, and the similar image is set so that the distance of the abscissa between the lower limit and the threshold is the same on the histogram of the reference image and the similar image. Set the threshold of. FIG.
(A) shows the histogram of the reference image, and (B) shows the histogram of the similar image. The lower limit (X1) of the histogram of the reference image and the lower limit (X1) of the histogram of the similar image
By adding the difference ΔX = X2−X1 to 2) to the abscissa of the threshold values T1 to T3 set in the histogram of the reference image, the threshold values T1 to T3 can be set on the histogram of the similar image. This setting is also performed by the CPU 6, the image processor 7, and the like.

Next, a fifth embodiment will be described. In this embodiment, the center of gravity is used as the feature position of the histogram of the reference image, the center of gravity and the center of gravity of the histogram of the similar image are matched by adjusting the illumination intensity on the measurement object, and the threshold of the histogram of the reference image is similar. It is used as a threshold value of an image histogram, and has the same principle as that of the first embodiment. First, the center of gravity X1 of the histogram of the reference image and the center of gravity X2 of the histogram of the similar image are obtained, and the lighting current of the lighting device 23 is controlled by the automatic lighting control device 17 shown in FIG. 1 so that the center of gravity X2 matches the center of gravity X1 (X2 = X1) to adjust the illumination intensity. After matching both centroids, the positions of the threshold values T1 to T3 set on the histogram of the reference image are set in the histogram of the similar image. The calculation of the centroids X1 and X2 and the determination of coincidence are performed by the image processor 7, and the control of the automatic lighting adjustment device 17 is performed by the control management computer 14. Although the center of gravity is used as the feature position, the threshold can be determined in the same manner when a median or lower limit of density is used instead of the center of gravity.

The five threshold value determination methods have been described above.
Which method is to be adopted depends on the object to be measured. However, it is not always possible to determine which one to use. In such a case, it is better to determine an appropriate method by trial and error. After determining the threshold value of each image as described above, the area of each region is calculated, and quantitative measurement of each tissue is performed.

In the above embodiment, an example in which the tissue is observed and identified by using a microscope has been described. However, not only the identification of the metal tissue and the biological tissue, but also the inspection of the printing result, the identification of the temperature region of the temperature distribution image, and the satellite It can also be used to measure ground area distribution using photographs and aerial photographs.

[0029]

As is clear from the above description, according to the present invention, if a threshold value is set on the histogram of the reference image, the similarity image of the similar image is determined based on the characteristics of the histogram of the reference image and the similar image. The threshold can be set automatically.
Thus, a threshold can be set even when a clear bottom does not appear in the histogram of a similar image.

[Brief description of the drawings]

FIG. 1 is a diagram showing an example of a configuration for realizing an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a metal surface structure.

FIG. 3 is a diagram showing a density histogram of the tissue shown in FIG. 2;

FIG. 4 is an operation flowchart of the embodiment of the present invention.

FIG. 5 is an operation flowchart following FIG. 4;

FIG. 6 is an explanatory diagram for setting a threshold value in the first embodiment.

FIG. 7 is a diagram illustrating a method of setting a threshold value by adjusting the center of gravity of the second embodiment.

FIG. 8 is a diagram illustrating a method of setting a threshold value by adjusting a median value according to the third embodiment.

FIG. 9 is a diagram illustrating a method of setting a threshold value by adjusting a lower limit value according to the fourth embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Microscope 6 CPU 7 Image processor 8 Grayscale image memory 9 Binary memory 15 Autostage device 16 Autofocus device 17 Automatic lighting adjustment device 20 Imaging device 23 Illumination device

Claims (5)

(57) [Claims] 1. A histogram of a grayscale distribution area is created for an image representing a plurality of regions according to a predetermined grayscale level,
In the method for determining a threshold value of an area by image analysis that defines a threshold value representing a density range corresponding to each target area with respect to the histogram, a characteristic position indicating a characteristic of a histogram of a reference image is determined, and the histogram is associated with each target area. A threshold representing the density range is determined, and a feature position of the histogram is obtained for a similar image similar to the reference image, and a difference between the feature position of the histogram of the reference image and the feature position of the histogram of the similar image is set as a threshold of the histogram of the reference image. A method for determining a threshold value of a region by image analysis, wherein the added position is used as a threshold value of a histogram of a similar image. 2. A histogram of a distribution area of shades is created in accordance with a preset shade level from an image representing a plurality of areas taken by a television camera by illuminating a sample, and the histogram corresponds to each target area. In a method of determining a threshold value of an area by image analysis that defines a threshold value representing a density range to be obtained, a histogram of a reference image and its characteristic position are obtained, a threshold value representing a density range corresponding to each target area is determined in the histogram, and a similarity to the reference image is determined. The histogram and the feature position of the similar image to be obtained are obtained, the illumination is adjusted so that the feature position of the histogram of the similar image matches the feature position of the histogram of the reference image, and the threshold of the histogram of the reference image is set as the histogram threshold of the similar image. A method for determining a threshold value of an area by image analysis. 3. A threshold determination method in the region by the image analysis of <br/> claim 1 or 2, characterized in that the characteristic position is the value of the concentration of the area center of gravity of the histogram. 4. The method according to claim 1, wherein the characteristic position is a median between an upper limit value and a lower limit value of the density of the histogram.
3. The method for determining a threshold value of an area by image analysis according to 1 or 2 . 5. The threshold determination process area by image analysis according to claim 1 or 2, characterized in that the characteristic position is the lower limit value of the density of the histogram.