JP2022166376A - Device and method for determining state of polishing and quality evaluation device - Google Patents

Abstract

To provide a technique that can properly determine the state of polishing of the surface of metal.SOLUTION: A quality evaluation device 4 has a processor 5 for performing: acquisition processing 5a of acquiring a target image to be a determination target from a microscopic image of the surface of metal corroded by a tissue observation corrosion liquid after polishing; image generation processing 5c of generating, from a grayscale image based on the target image, at least one of a bright-side image, in which particles of tissues of the metal are shown as bright objects and the others are shown as dark regions, and a dark-side image, in which the particles of tissues of the metal are shown as dark objects and the others are shown as bright regions; polishing state determination processing 5d of outputting the result of determination as to the polishing state of the surface of the metal on the basis of a value that shows the degree of flatness of the objects in the image generated by the image generation processing 5c.SELECTED DRAWING: Figure 1

Description

The present invention relates to a polishing state determination device, a polishing state determination method, and a quality evaluation device.

Metal materials, particularly steel materials, are sometimes evaluated for their quality based on physical properties such as carbon concentration, retained austenite amount, and hardness.

For example, the amount of retained austenite is measured by X-ray analysis. Patent Literature 1 discloses a technique for determining the amount of retained austenite in steel materials based on the results of X-ray analysis.

JP 2018-040770 A

By the way, as a quality evaluation of a metal material, an evaluation by microscopic observation of its structure is also performed.
The metallographic structure reflects the composition of the metal material and the history of heat treatment. Therefore, it is conceivable to obtain estimated values of physical properties such as carbon concentration, retained austenite amount, and hardness of the metal material by importing the microscopic image of the metal structure into a computer and quantifying the characteristics of the metal structure.

The operation of picking up a microscopic image of the metal structure includes operations such as metal surface polishing, corrosion, and microscopic observation. Since work such as polishing, corrosion, and microscopic observation of metal surfaces is performed manually by workers, the state of polishing and corrosion of metal surfaces can be good or bad. Possible.

Here, when trying to obtain an estimated physical property value of a metal material based on a microscopic image, if the polished state or corrosion state of the metal surface captured as the microscopic image is in a poor state, the physical property value There is a risk that the estimation accuracy of
Therefore, it is necessary to appropriately determine the polished state and corroded state of the metal surface captured as a microscope image.

The polishing state determination device, which is an embodiment,
Acquisition processing for acquiring a target image to be determined from a microscopic image of a metal surface that has been corroded by a etchant for structure observation after polishing;
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. an image generation process that generates at least one of the dark side images shown as regions;
polishing state determination processing for outputting a determination result regarding the polishing state of the metal surface based on the value indicating the flatness of the object included in the image generated by the image generation processing;
A processing unit for executing

A polishing state determination method, which is an embodiment viewed from another point of view, comprises:
A polished state determination method for determining the polished state of a metal surface corroded by a corrosive liquid for structure observation after polishing, comprising:
Acquiring a target image to be determined from the microscope image of the metal surface,
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. generating at least one of the dark side images shown as regions;
Based on the value indicating the flatness of the object included in the generated image, a determination result regarding the polished state of the metal surface is output.

In addition, the quality evaluation device, which is an embodiment viewed from another point of view,
A quality evaluation device for evaluating the quality of the metal based on a microscope image of the metal surface corroded by the etchant for structure observation after polishing,
Acquisition processing for acquiring a target image from the microscope image, which is an object for determining the polishing state of the metal surface;
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. an image generation process that generates at least one of the dark side images shown as regions;
polishing state determination processing for outputting a determination result regarding the polishing state of the metal surface based on the value indicating the flatness of the object included in the image generated by the image generation processing;
When the judgment result of the polishing state judgment processing satisfies a predetermined condition, an area image of an area to be subjected to quality evaluation is obtained from the microscope image, and the light side image and the bright side image and the light side image are obtained from a grayscale image based on the area image. a quality evaluation process of generating a dark side image and obtaining an estimated physical property value of the metal based on the bright side image and the dark side image;
A processing unit for executing

According to the present disclosure, it is possible to appropriately determine the polished state of the metal surface.

FIG. 1 is a block diagram showing an example of a quality evaluation system. FIG. 2 is a flow chart showing an example of a quality evaluation method using a quality evaluation system. FIG. 3 is a diagram for explaining the steps of preparing a sample, polishing the sample, and corroding the sample. FIG. 4 is a flowchart showing an example of quality evaluation processing. FIG. 5A is an example of an image showing the entire imaging target range, and FIG. 5B is an example of an area image obtained from the image showing the entire imaging target range. FIG. 6(a) is a diagram showing a part of the area image, FIG. 6(b) is a diagram of a dark side image generated from this area image, and FIG. 6(c) is a diagram generated from this area image. FIG. 10 is a diagram of a bright-side image; FIG. 7 is a diagram showing an example of a histogram based on the brightness of each pixel forming an area image and its appearance frequency. FIG. 8 is an explanatory diagram showing one of the dark objects included in the dark side image (FIG. 6(b)). FIG. 9 is an explanatory diagram showing one bright object included in the bright side image (FIG. 6(c)). FIG. 10 is a flowchart illustrating an example of determination processing. 11A and 11B are diagrams showing target images of cross sections obtained by mirror-polishing a sample produced using an evaluation target member and then corroding the sample with different corrosion times. FIG. 12 is a diagram showing histograms based on luminance values (brightness) of each of these target images. FIG. 13 is a diagram showing the relationship between the mode of each histogram and the difference between the estimated carbon concentration value and the actual measurement value of the target image corresponding to each histogram. FIG. 14 is a diagram showing a target image of a cross section of a sample produced using the member to be evaluated, which is corroded after being polished under different polishing conditions. FIG. 15 is a diagram showing the relationship between the aspect ratio of an object included in the light-side image and the difference between the estimated carbon concentration value and the actual measurement value for each target image. FIG. 16 is a diagram showing part of a flowchart of determination processing according to a modification.

First, the contents of the embodiments will be listed and explained.
[Overview of embodiment]
(1) A polishing state determination device according to an embodiment includes:
Acquisition processing for acquiring a target image to be determined from a microscopic image of a metal surface that has been corroded by a etchant for structure observation after polishing;
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. an image generation process that generates at least one of the dark side images shown as regions;
polishing state determination processing for outputting a determination result regarding the polishing state of the metal surface based on the value indicating the flatness of the object included in the image generated by the image generation processing;
A processing unit for executing

If there are polishing marks on the metal surface, the polishing marks appear as objects in the image generated by the image generation process. Since the polishing marks are linear, the flatness of the object due to the polishing marks is greater than the flatness of the object of the tissue grains.
If there are many objects with polishing marks among the objects included in the image generated by the image generation processing, the value indicating the flatness of the object becomes relatively large. That is, the value indicating the flatness of the object indicates the amount of polishing marks.
Therefore, according to the above configuration, the judgment result regarding the polished state of the metal surface is output based on the value indicating the flatness of the object affected by the presence or absence of polishing marks during polishing. can be determined accurately. As a result, the polished state of the metal surface can be determined appropriately.

(2) In the polishing state determination device,
It is preferable that the value indicating the flatness of the object includes at least one of the aspect ratio, circularity, and major axis length of the object.
In this case, the polished state of the metal surface can be quantitatively determined using at least one of the aspect ratio, circularity, and major axis length of the object.

(3) In the polishing state determination device,
The processing unit determines a corrosion state of the metal surface, prior to the image generation processing, based on a histogram based on a luminance value representing brightness of each pixel constituting the target image and its appearance frequency. Execute the judgment process,
In the image generation process, at least one of the bright side image and the dark side image may be generated when the determination result of the corrosion state determination process satisfies a predetermined condition.
In this case, after quantitatively determining the corrosion state of the metal surface, the polishing state of the metal surface can be quantitatively determined.

(4) A polishing state determination method, which is an embodiment viewed from another point of view,
A polished state determination method for determining the polished state of a metal surface corroded by a corrosive liquid for structure observation after polishing, comprising:
Acquiring a target image to be determined from the microscope image of the metal surface,
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. generating at least one of the dark side images shown as regions;
Based on the value indicating the flatness of the object included in the generated image, a determination result regarding the polished state of the metal surface is output.

According to the above configuration, the determination result regarding the polished state of the metal surface is output based on the value indicating the flatness of the object affected by the presence or absence of polishing marks during polishing, so the polished state of the metal surface can be appropriately determined. can do.

(5) In addition, the quality evaluation device, which is an embodiment viewed from another point of view,
A quality evaluation device for evaluating the quality of the metal based on a microscope image of the metal surface corroded by the etchant for structure observation after polishing,
Acquisition processing for acquiring a target image from the microscope image, which is an object for determining the polishing state of the metal surface;
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. an image generation process that generates at least one of the dark side images shown as regions;
polishing state determination processing for outputting a determination result regarding the polishing state of the metal surface based on the value indicating the flatness of the object included in the image generated by the image generation processing;
When the judgment result of the polishing state judgment processing satisfies a predetermined condition, an area image of an area to be subjected to quality evaluation is obtained from the microscope image, and the light side image and the bright side image and the light side image are obtained from a grayscale image based on the area image. a quality evaluation process of generating a dark side image and obtaining an estimated physical property value of the metal based on the bright side image and the dark side image;
A processing unit for executing

According to the above configuration, the polished state of the metal surface can be determined quantitatively, so that the polished state of the metal surface can be appropriately determined. As a result, for example, a microscopic image of a poorly polished metal surface can be prevented from being used in the quality evaluation process, thereby suppressing a decrease in accuracy in estimating physical property values.

[Details of embodiment]
Preferred embodiments are described below with reference to the drawings.
[Regarding the overall system configuration]
FIG. 1 is a block diagram showing an example of a quality evaluation system.
A quality evaluation system 1 in FIG. 1 has a function of evaluating the quality of metal based on a microscope image of the metal surface.
A quality evaluation system 1 includes a microscope 2 , a camera 3 and a quality evaluation device 4 .
The microscope 2 is a metallurgical microscope capable of observing metal surfaces.
The camera 3 is incorporated in the microscope 2 and has a function of capturing an observation image of the microscope 2 . The camera 3 is connected to the quality evaluation device 4, and provides the captured microscope image to the quality evaluation device 4 as image data.

The quality evaluation device 4 is composed of a computer having a processor 5 and a storage device 6 . The quality evaluation device 4 also includes an input/output unit 7 having a function of receiving input from the operator and outputting information and the like to the operator. The input/output unit 7 is, for example, a keyboard, mouse, touch panel, monitor, speaker, printer, indicator, and the like.

The storage device 6 stores computer programs and the like executed by the processor 5 . The processor 5 reads the computer program recorded in a computer-readable non-transitory recording medium such as the storage device 6, and executes each process of the processor 5 described below.
The storage device 6 also stores a microscope image 6a provided from the camera 3, threshold data 6b, and a quality evaluation model 6c.
A microscope image 6 a is image data provided from the camera 3 .
The threshold data 6b includes upper and lower limit values Th1 to Th4, which are threshold values used in the corrosion state determination process and the polishing state determination process, which will be described later. The processor 5 refers to the upper and lower limit values Th1 to Th4 of the threshold data 6b as necessary.
The quality evaluation model 6c is a model used in quality evaluation processing, which will be described later. The quality evaluation model 6c includes a carbon concentration estimation model, a retained austenite amount estimation model, and a hardness estimation model.

The processor 5 executes the acquisition process 5a, the corrosion state determination process 5b, the image generation process 5c, the polishing state determination process 5d, and the quality evaluation process 5e by executing the computer program described above. Each of these processes will be described later.

[About the quality evaluation method]
FIG. 2 is a flow chart showing an example of a quality evaluation method using the quality evaluation system 1. FIG.
The system 1 can evaluate the quality of heat-treated members made of low-carbon steel, low-carbon alloy steel, or the like. Low carbon steels include S10C, S25C, and S40C. SCr415, SCM415, SCM435, SNCM420, SAE5120 etc. are mentioned as low carbon alloy steel.

In the following, a description will be given of a case where a bearing ring of a rolling bearing is used as an evaluation object member that is an object of quality evaluation. This bearing ring is a member formed using low-carbon alloy steel or low-carbon steel, more specifically SAE5120. The bearing rings are tempered after carburizing and quenching.
In this quality evaluation method, estimated values of physical properties such as carbon concentration, amount of retained austenite, and hardness of the carburized portion of the bearing ring are obtained as values used for quality evaluation. Therefore, the carburized layer of the bearing ring is an area subject to quality evaluation.

In this quality evaluation method, first, a bearing ring is used to prepare a sample (Step S1 in FIG. 2), and the sample is polished and corroded (Step S2 in FIG. 2).

FIG. 3 is a diagram for explaining the steps of preparing a sample, polishing the sample, and corroding the sample.
In the sample preparation process, first, a cut piece 102 is obtained by cutting the bearing ring 100, which is a member to be evaluated, with a high-speed cutter or the like. A sample 106 is obtained by embedding the cut piece 102 in a resin 104 .
At this time, the cross section 102 a of the cut piece 102 is exposed on the observation surface 106 a of the sample 106 .
Next, in the polishing and corrosion process, the observation surface 106a of the sample 106 is polished with a metal polisher or the like. The observation surface 106a is polished with diamond paste or the like after being polished with waterproof paper. Thereby, the observation surface 106a (cross section 102a) is mirror-polished.

Next, the mirror-polished observation surface 106a is immersed in an etchant for structure observation such as a nital solution (nitric acid alcohol solution) for a predetermined period of time to corrode (etch) the cross section 102a.

Next, the sample 106 is set on the microscope 2, a necessary portion of the cross section 102a is set in the field of view of the microscope 2, and a microscope image is captured (step S3 in FIG. 2).
In this method, the carburized layer of bearing ring 100, which is the member to be evaluated, is subject to quality evaluation. Thereby, a microscope image of the carburized layer in the cross section 102a is captured.
The work from steps S1 to S3 is performed by an operator.

A microscope image captured by camera 3 is provided to quality evaluation device 4 .
The quality evaluation device 4 provided with the microscope image executes determination processing (step S4 in FIG. 2).
The determination process in step S4 in FIG. 2 determines the corroded state and polished state of the cross section 102a of the sample 106, which is the metal surface corroded by the etchant for structure observation, and determines the corroded state and polished state of the cross section 102a. is output.
This determination processing will be described later.

The quality evaluation device 4 outputs to the operator information indicating either "accepted" or "failed" as the judgment result regarding the corroded state and polished state of the cross section 102a.
When the result of the determination process output from the quality evaluation device 4 indicates "failed", the operator polishes and corrodes the sample again (step S2 in FIG. 2), and repeats the work sequentially.

On the other hand, when the result of the determination process output from the quality evaluation device 4 indicates "accepted", the operator picks up a microscope image of the sample 106 (step S6 in FIG. 2).
The microscope image captured here is an image used for quality evaluation processing. The quality evaluation process may require an image of the metal structure in a wider range than the imaging range of the microscope image. In this case, in step S6, a range wider than the imaging range of the microscope image is set as the imaging target range. This imaging target range is divided into a plurality of areas according to the imaging range of the microscope image, and a plurality of microscope images obtained by imaging the divided areas are obtained. More specifically, the carburized layer exists in a range from the raceway surface 102a1 (FIG. 3) in the cross section 102a to a depth of about 1 mm to 3 mm directly below the raceway surface 102a1. Therefore, an imaging target range is determined along the contour of the track surface 102a1 so as to include the carburized layer, and the imaging target range is divided along the contour of the track surface 102a1 to obtain a plurality of microscope images.

In FIG. 2, the microscopic image captured in step S6 is given to the quality evaluation device 4. As shown in FIG.
The processor 5 of the quality evaluation device 4 to which the multiple microscopic images have been supplied stores the multiple microscopic images in the storage device 6 and executes the quality evaluation processing 5e (step S7 in FIGS. 1 and 2).

FIG. 4 is a flowchart showing an example of quality evaluation processing.
When an instruction to perform quality evaluation processing is given along with a plurality of microscope images, the processor 5 starts the quality evaluation processing 5e and acquires an area image of the area subject to quality evaluation (step S10 in FIG. 4). ).
An area image is obtained from the given microscopic images. The processor 5 connects a plurality of microscope images to generate an image showing the entire imaging target range. The processor 5 acquires an image of the carburized layer portion in the cross section 102a as an area image from the image showing the entire imaging target range.

FIG. 5A is an example of an image showing the entire imaging target range, and FIG. 5B is an example of an area image obtained from the image showing the entire imaging target range.
In FIG. 5A, the cross section 102a is imaged in a range of several hundred μm in the depth direction from the raceway surface 102a1 along the contour of the raceway surface 102a1.
For example, the processor 5 acquires a belt-like area near a point 50 to 100 μm in the depth direction from the track surface 102a1 in FIG. 5(a) as an area image. The width dimension of the area image in the depth direction is about 50 to 100 μm.

After obtaining the area image, the processor 5 converts the area image into a grayscale image and then performs image processing (step S11 in FIG. 4).
Image processing includes preprocessing and processing to generate a bright side image and a dark side image.
The preprocessing includes a process of removing noise contained in the area image converted to the grayscale image, and a process of clarifying the contours of the crystal grains captured in the area image.

Since the area image is obtained by connecting a plurality of microscope images, the brightness value (brightness) varies periodically. In the process of removing noise, such noise that appears as periodic fluctuations in brightness is removed by correcting the brightness of each pixel that constitutes the area image.
In addition, in the process of clarifying the outline of the crystal grain, the brightness of each pixel constituting the area image and the brightness of each pixel are corrected so that the slope of the cumulative frequency of the histogram based on the frequency of appearance thereof becomes constant. .

After finishing the image processing, the processor 5 executes processing for generating a bright side image and a dark side image.
FIG. 6(a) is a diagram showing a part of the area image, FIG. 6(b) is a diagram of a dark side image generated from this area image, and FIG. It is a figure of a bright side image.
The dark side image is an image in which tissue grains in the area image are shown as dark objects (black objects) and other areas are shown as bright areas (white areas).
The bright-side image is an image in which tissue grains in the area image are shown as bright objects (white objects) and other areas are shown as dark areas (black areas).

The grain shown as a dark object in the dark side image (FIG. 6B) and the grain shown as a bright object in the bright side image (FIG. 6C) indicate different grains.
Dark objects (dark areas) are areas where corrosion due to the etching has progressed, and bright objects (bright areas) are areas where corrosion due to the etching has not progressed.
The bright-side and dark-side images are generated by processor 5 as follows.
First, a histogram is generated from the preprocessed grayscale image (FIG. 6(a)). This histogram is a histogram based on the brightness of each pixel forming the area image and its appearance frequency.
FIG. 7 is a diagram showing an example of the above histogram, in which the horizontal axis is brightness (0 to 255) and the vertical axis is appearance frequency. FIG. 7 also shows a curve L indicating the cumulative frequency for each brightness (for each gradation).

A first threshold and a second threshold are set in advance for the lightness in order to generate the bright side image and the dark side image.
In the histogram shown in FIG. 7, the lightness at which the cumulative ratio from the dark side (lightness 0) is the first set value is set as the first threshold. For example, a value of 30% or more and 50% or less is adopted as the first set value. For example, when 40% is adopted as the first set value, in the histogram shown in FIG. 7, the brightness (here, the brightness is "100") at which the cumulative ratio from the dark side (brightness 0) is 40%. is set as the first threshold.
Also, in the histogram shown in FIG. 7, the lightness at which the cumulative ratio from the bright side (lightness 255) is the second set value is set as the second threshold. For example, a value of 10% or more and 30% or less is adopted as the second set value. For example, when 20% is adopted as the second set value, in the histogram shown in FIG. 7, the lightness (here, lightness "180") at which the cumulative ratio from the bright side (lightness 255) is 20%. is set as the second threshold. The second threshold becomes a larger value than the first threshold.

Among the pixels of the preprocessed area image (FIG. 6(a)), the pixels whose lightness is equal to or less than the first threshold value (lightness is equal to or less than 100) are regarded as elements forming a dark object. On the other hand, a pixel whose lightness exceeds the second threshold (lightness exceeds 180) is regarded as an element forming a bright object. As a result, a dark side image (FIG. 6B) is generated from the area image (FIG. 6A) in which the pixels below the first threshold are black and the other areas are white. Further, from the area image (FIG. 6(a)), a bright side image (FIG. 6(c)) is generated in which the area exceeding the second threshold is white and the other area is black.
As described above, the processor 5 generates the bright side image and the dark side image based on the area image.

After generating the bright-side image and the dark-side image, the processor 5 calculates feature amounts based on the bright-side image and the dark-side image (step S12 in FIG. 4).
As shown in FIG. 6B, the dark side image is an image in which tissue grains are shown as dark objects (black objects). The processor 5 obtains dark side feature amounts of a plurality of objects included in the dark side image (FIG. 6(b)).
As shown in FIG. 6C, the bright side image is an image in which tissue grains are shown as bright objects (white objects). The processor 5 obtains bright-side feature amounts of a plurality of objects included in the bright-side image (FIG. 6(c)).

FIG. 8 is an explanatory diagram showing one of the dark objects included in the dark side image (FIG. 6(b)). FIG. 9 is an explanatory diagram showing one bright object included in the bright side image (FIG. 6(c)). A plurality of objects are extracted from each of the bright side image and the dark side image. The object to be extracted is one block part with a given area. An object whose area is smaller than the threshold and an object whose area is larger than the threshold are excluded from the objects to be extracted. A dark side feature amount of each object extracted from the dark side image is obtained. A bright side feature amount of each object extracted from the bright side image is obtained.

Examples of the dark side feature amount include long axis, short axis, angle, degree of circularity, maximum value of lightness, degree of skewness, number of pixels, aspect ratio, degree of tone, etc. of a dark object.
Examples of the light-side feature amount include the maximum brightness value, major axis, angle, circularity, median brightness value, kurtosis, skewness, number of pixels, and aspect ratio of a bright object.
As shown in FIGS. 8 and 9, the major axis is the maximum dimension of the object and the minor axis is the maximum dimension perpendicular to the major axis. The angle is the inclination angle of the long axis with respect to the width direction. The degree of circularity is a calculated value of “4π×(area of object)/(perimeter of object)̂2”. Aspect ratio is a calculated value of "major axis/minor axis". The degree of alignment is a calculated value of "(area of object)/(area of smallest rectangle enclosing object)". The median values of brightness are as follows. "A median value of a histogram created with brightness on the horizontal axis and frequency on the vertical axis for an object for which a feature amount is to be obtained in a grayscale image (FIG. 6(a))". The maximum brightness values are as follows. "Maximum value of the histogram if the above histogram is created". Skewness and kurtosis are values calculated from the shape of the histogram.

After calculating the bright side feature amount and the dark side feature amount based on the bright side image and the dark side image, the processor 5 normalizes the bright side feature amount and the dark side feature amount of each of the plurality of objects. Estimated physical property values such as carbon concentration, retained austenite amount, and hardness are calculated based on the bright side feature amount and the dark side feature amount (step S13 in FIG. 4).
The processor 5 refers to the quality evaluation model 6c in the storage device 6 and calculates the estimated physical property value.

The quality evaluation model 6c includes the carbon concentration estimation model, the retained austenite amount estimation model, and the hardness estimation model, as described above.
Each of these models is an arithmetic expression for obtaining an estimated value of carbon concentration, an estimated amount of retained austenite, and an estimated value of hardness based on the bright side feature amount and the dark side feature amount. Each model is machine-learned using teacher data in which the bright-side feature amount, the dark-side feature amount, and each estimated value are associated with each other.
The processor 5 uses the bright side feature amount and the dark side feature amount obtained in step S12 in FIG. 4 as a data set, calculates each estimated value, outputs the estimated value through the input/output unit 7, and ends the quality evaluation process.

[Regarding judgment processing]
In FIG. 2, the quality evaluation device 4 to which the microscopic image captured in step S3 is given stores the microscopic image in the storage device 6, and executes determination processing (step S4 in FIG. 2).
FIG. 10 is a flowchart illustrating an example of determination processing.
As described above, the determination process determines the corroded state and polished state of the cross section 102a of the sample 106, which is the metal surface corroded by the etchant for structure observation, and outputs the determination results regarding the corroded state and polished state of the cross section 102a. It is a process to

In the determination process, the processor 5 first executes the acquisition process 5a (FIG. 1) to acquire a target image to be determined from the given microscope image (step S21 in FIG. 10).
The target image is an image for which the corroded state and polished state of the cross section 102a are to be determined.
The target image may be an image of the portion of the cross section 102a included in the microscope image, but it is preferable to acquire the image of the carburized portion of the cross section 102a as the target image, as with the region image in step S10 in FIG.

After obtaining the target image, the processor 5 executes the corrosion state determination processing 5b (FIG. 1), and determines the brightness value (brightness or darkness) representing the brightness of each pixel constituting the target image and the frequency of appearance thereof. A judgment value is obtained from the histogram based on the above, and the obtained judgment value is compared with a threshold value set for the judgment value. In this embodiment, the mode value that can be obtained from the histogram is used as the determination value (step S22 in FIG. 10). value thresholds (upper limit value Th1, lower limit value Th2) are compared (step S23).
More specifically, the processor 5 determines whether or not the mode is equal to or greater than the lower limit value Th2 and equal to or less than the upper limit value Th1 (step S23).

The upper limit value Th1 and the lower limit value Th2 set for the mode value are set according to the material of the member to be evaluated. For example, in the case of the material (low-carbon alloy steel or low-carbon steel) of the member to be evaluated in this embodiment, the upper limit value Th1 is set to 185 and the lower limit value Th2 is set to 70.
These upper limit value Th1 and lower limit value Th2 are set as follows.

That is, a plurality of target images of the cross section 102a with different corrosion states were acquired, and the mode of the histogram based on the luminance value of each target image was obtained. Furthermore, the estimated value of the carbon concentration is obtained using each target image, the difference between the measured value and the estimated value of carbon concentration is obtained for each target image, the difference of each target image, and the histogram corresponding to each target image. The upper limit value Th1 and the lower limit value Th2 were set based on the mode and the frequency.

FIG. 11 is a diagram showing a target image of a cross section 102a of a sample 106 produced using a member to be evaluated, which is mirror-polished and then corroded for different corrosion times.
Each target image is an image of a cross section 102a obtained by mirror-polishing the same sample 106 under the same conditions and corroding it for different corrosion times. In addition, a 3% nital solution was used for corrosion, and the target image was acquired with the brightness of the light source of the microscope 2 being constant when acquiring each image.
In FIG. 11, target images with corrosion times of 1 second, 3 seconds, 5 seconds, 7 seconds, and 10 seconds are shown in order from the left side of the paper. As shown in FIG. 11, the corrosion of the cross section 102a progresses as the corrosion time increases, and the texture grains in the cross section 102a appear densely.

FIG. 12 is a diagram showing histograms based on luminance values (brightness) of each of these target images. In FIG. 12, the horizontal axis represents brightness, which is the brightness value of each pixel forming the target image. The vertical axis indicates the appearance frequency of each brightness. Assume that the lightness has 256 gradations from 0 to 255, as described above.
In FIG. 12, the histogram H1 is the histogram corresponding to the target image with the corrosion time of 1 second, the histogram H3 is the histogram corresponding to the target image with the corrosion time of 3 seconds, and the histogram H5 is the histogram corresponding to the target image with the corrosion time of 5 seconds. H7 is the histogram corresponding to the target image with the corrosion time of 7 seconds, and histogram H10 is the histogram corresponding to the target image with the corrosion time of 10 seconds.
In FIG. 12, the mode of histogram H1 is 189, the mode of histogram H3 is 183, the mode of histogram H5 is 133, the mode of histogram H7 is 73, and the mode of histogram H10 is 23. .

In this way, a plurality of target images of the cross section 102a with different corrosion states (target images with different histogram modes) are acquired, and quality evaluation processing 5e (FIG. 1) is performed using each target image. An estimate of the carbon concentration was obtained from the target image.
Next, the carbon concentration is actually measured by quantitative analysis by EPMA on the cross section 102a, the difference between the actual measurement value and each estimated value is obtained, and for each histogram (each target image), the difference between the actual measurement value and the estimated value and the maximum The upper limit value Th1 and the lower limit value Th2 for the mode value are set. Note that the measured value of the carbon concentration of (the carburized portion of) the cross section 102a of the target image shown in FIG. 12 was 0.8.

FIG. 13 is a diagram showing the relationship between the mode of each histogram and the difference between the estimated carbon concentration value and the actual measurement value of the target image corresponding to each histogram. In FIG. 13, the horizontal axis indicates the mode of the histogram. The vertical axis represents the ratio (%) of the difference between the estimated value and the measured value of the carbon concentration to the measured value.

In FIG. 13, a group of black circles P1 with a mode value of 189 indicates the difference due to the target image corresponding to the histogram H1. A group of black circles P3 whose mode is 183 indicates the difference due to the target image corresponding to the histogram H3. A group of black circles P5 with a mode value of 133 indicates the difference due to the target image corresponding to the histogram H5. A group of black circles P7 whose mode is 73 indicates the difference due to the target image corresponding to the histogram H7. A group of black circles P10 with a mode value of 23 indicates the difference due to the target image corresponding to the histogram H10.

As shown in FIG. 13, the difference between the target images corresponding to the histograms H3, H5 and H7 is within ±0.1%, while the difference between the target images corresponding to the histograms H1 and H10 is ±0.1%. exceeds
Based on this result, the upper limit Th1 is set to 185 and the lower limit Th2 is set to 70 for the mode.
As a result, if the mode of the histogram based on the luminance value of the target image is equal to or greater than the lower limit Th2 and equal to or less than the upper limit Th1, the difference between the estimated carbon concentration value of the target image and the measured value is within ±0.1%. becomes.

If it is determined in step S23 in FIG. 10 that the mode value of the histogram based on the luminance value of the target image acquired in step S21 in FIG. The processor 5 executing the process 5b (FIG. 1) proceeds to step S25 in FIG. 10, outputs information indicating "failed" to the operator via the input/output unit 7, and finishes the determination process.
In this case, since the judgment result is "fail", the operator again polishes and corrodes the sample (step S2 in FIG. 2) as described in step S5 in FIG. Start over.

On the other hand, when determining that the mode of the histogram based on the luminance values of the target image acquired in step S21 in FIG. move on.
In this case, the processor 5 determines that the corrosion state of the cross section 102a of the target image obtained in step S21 in FIG. 10 is "accepted". If the determination result in step S27 in FIG. 10 is affirmative, the processor 5 outputs information indicating "passed" to the operator.

In step S23 in FIG. 10, the processor 5 of the present embodiment outputs information indicating "pass" or information indicating "failure" to the operator according to the mode value as the judgment value. In other words, in the corrosion state determination process 5b, the processor 5 determines the cross section 102a based on the mode value (determined value) obtained from the histogram of the luminance values representing the brightness of each pixel constituting the target image and the appearance frequency thereof. output the judgment result about the corrosion state of

When the cross-section 102a is corroded by the etchant for observing the structure, the cross-section 102a has a brightness (darkness) contrast corresponding to the structure. Therefore, the luminance value of each pixel forming the target image is affected by the state of corrosion.
Therefore, according to the above configuration, since the corrosion state of the cross section 102a is determined based on the mode value obtained from the histogram based on the luminance values affected by the corrosion state, the corrosion state of the cross section 102a can be quantitatively determined. The determination result can be output. As a result, the corrosion state of the cross section 102a can be appropriately determined.

In addition, the brightness of the light source of the microscope 2 may also affect the brightness value of each pixel forming the target image. can judge.

In the above configuration, the corrosion state of the cross section 102a is determined by comparing the mode obtained from the histogram with the upper limit value Th1 and the lower limit value Th2. Determination of the corrosive state can be performed appropriately. That is, if the upper limit value Th1 and the lower limit value Th2 are set so that the mode value is suitable for the quality evaluation process 5e (FIG. 1), it is possible to determine whether or not the corrosion state is suitable for the quality evaluation process 5e. can be done.

Proceeding to step S24 in FIG. 10, the processor 5 executes the image generation process 5c (FIG. 1) to generate a bright side image based on the target image acquired in step S21 in FIG. middle, step S24).
The bright side image is the same image as the bright side image generated in the image processing of step S11 in FIG.
The processor 5 generates a bright side image by the same processing as the image processing in step S11 in FIG. 4 (step S24 in FIG. 10).
In this embodiment, a bright side image is generated in step S24 in FIG. 10, but a dark side image may be generated instead of the bright side image.

After generating the light-side image, the processor 5 executes a polishing state determination process 5d to obtain an aspect ratio as a value indicating the flatness of a plurality of objects included in the light-side image generated by the image generation process 5c (step S26). The processor 5, for example, normalizes the aspect ratio of each of the plurality of objects and processes using the normalized aspect ratio.
The processor 5 compares the obtained aspect ratio with the upper limit set for the aspect ratio (step S27).
More specifically, the processor 5 determines whether the aspect ratio is equal to or less than the upper limit Th3 (step S27).

The upper limit value Th3, which is the flatness threshold value set for the aspect ratio, is set as follows.
That is, a plurality of target images of the cross section 102a with different polishing states were obtained, and the aspect ratio of the object included in the bright side image of each target image was obtained. Furthermore, the estimated value of the carbon concentration is obtained using each target image, and the difference between the measured value and the estimated value of the carbon concentration is obtained for each target image. The upper limit Th3 was set based on the aspect ratio of the objects involved.

FIG. 14 is a diagram showing a target image of a cross section 102a of a sample 106 produced using an evaluation target member, which is corroded after being polished under different polishing conditions.
Each target image G1, G2, G3, G4, and G5 is obtained by polishing the sample 106 under the conditions shown below, immersing the cross section 102a in a 3% nital solution for 5 seconds, and obtaining the brightness of the light source of the microscope 2 at a constant level. This is the target image that has been processed.

Polishing condition Target image G1: Polishing with waterproof paper (No. 220 10 seconds, No. 400 10 seconds, No. 800 10 seconds, No. 2000 10 seconds) → diamond polishing 30 seconds

Target image G2: Polishing with waterproof paper (No. 220 10 seconds, No. 400 10 seconds, No. 800 10 seconds, No. 2000 10 seconds) → Diamond polishing 120 seconds → Waterproof paper No. 2000 10 seconds → Diamond polishing 120 seconds

Target image G3: Polishing with waterproof paper (No. 220 10 seconds, No. 400 10 seconds, No. 800 10 seconds, No. 2000 10 seconds) → diamond polishing 10 seconds

Target image G4: Polishing with waterproof paper (No. 220 10 seconds, No. 400 10 seconds, No. 800 10 seconds, No. 2000 10 seconds)

Target image G5: Polishing with waterproof paper (No. 220 10 seconds, No. 400 10 seconds, No. 800 10 seconds, No. 2000 3 seconds)

The polishing conditions for the target image G1 are normal polishing conditions. The polishing conditions for the target image G2 are diamond polishing, polishing with water-resistant paper No. 2000, and diamond polishing to intentionally create minute hole-like defects called pits in the cross section 102a. . The polishing conditions for the target image G3 shorten the diamond polishing time compared to the normal polishing conditions. The polishing conditions for the target image G4 omit diamond polishing from the normal polishing conditions. As for the polishing conditions of the target image G5, the diamond polishing is omitted from the normal polishing conditions, and the polishing time with waterproof paper No. 2000 is shortened. Target images G3, G4, and G5 are conditions that reproduce a state in which polishing is insufficient.

Here, the polished state is the property of the polished surface due to polishing marks and the like, and cross sections 102a in different polished states can be obtained according to the polishing conditions described above.
As described above, a plurality of target images of the cross section 102a with different polishing states were acquired, and the aspect ratios of the plurality of objects included in the bright side image of each target image were obtained.
Furthermore, the quality evaluation process 5e (FIG. 1) was performed using each target image, and the estimated value of the carbon concentration by each target image was obtained.
Next, the difference between the actual measurement value and each estimated value of the carbon concentration was obtained, and for each target image, the difference between the actual measurement value and the estimated value was associated with the aspect ratio, and the upper limit value Th3 for the aspect ratio was set.

FIG. 15 is a diagram showing the relationship between the aspect ratio of an object included in the light-side image and the difference between the estimated carbon concentration value and the actual measurement value for each target image. In FIG. 15, the horizontal axis indicates the aspect ratio of the object included in the bright side image. The vertical axis represents the ratio of the difference between the estimated carbon concentration and the measured value to the measured value.

In FIG. 15, a group of black circles P10 indicates the aspect ratio and difference due to the target image G5. A group of black circles P11 indicates the aspect ratio and difference due to the target image G4. A group of black circles P12 indicates aspect ratios and differences between the target images G1, G2, and G3.

As shown in FIG. 15, the aspect ratio of the target image G5 is approximately four. Moreover, the difference between the actual measurement value and the estimated value of the carbon concentration in the target image G5 is 0.1% or more.
The aspect ratio of the target image G4 is approximately three. Moreover, the difference between the measured value and the estimated value of the carbon concentration in the target image G4 is 0.1% or more.
Also, the aspect ratio of the target images G1, G2, and G3 is approximately two. Moreover, the difference between the measured value and the estimated value of the carbon concentration in the target images G1, G2, and G3 is 0.05% or less.

Based on this result, the upper limit value Th3 for the aspect ratio of objects included in the bright side image is set to 2.5. That is, the allowable range of aspect ratio is set within 2.5.
As a result, if the aspect ratio of the object included in the target image is equal to or less than the upper limit value Th3, the difference between the estimated value of the carbon concentration in the target image and the measured value is 0.05% or less.

When it is determined in step S27 in FIG. 10 that the aspect ratio of the object included in the target image acquired in step S21 in FIG. Information indicating "failed" is output to the operator via the output unit 7, and the determination process is finished.
In this case, since the judgment result is "fail", the operator again polishes and corrodes the sample (step S2 in FIG. 2) as described in step S5 in FIG. Start over.

On the other hand, when determining that the aspect ratio of the object included in the target image is equal to or lower than the upper limit Th3, the processor 5 proceeds to step S28 in FIG. output to the user and finish the determination process.
In this case, since the determination result is "accepted", as described in step S5 in FIG. 2, the operator takes a microscope image for quality evaluation processing (step S6 in FIG. The processor 5 of the quality evaluation device 4 provided with the image executes a quality evaluation process 5e (step S7 in FIGS. 1 and 2).

As described above, the processor 5 performs determination processing to determine the corroded state and the polished state of the cross section 102a, and provides information indicating "accepted" or "failed" for the corroded state and the polished state of the cross section 102a. Output the information shown to the operator.
As a result, if the corroded state and polished state of the cross section 102a are good, the quality evaluation process 5e is executed, and if the corroded state and polished state of the cross section 102a are not good, the information indicating "failed" is displayed. By outputting, it is possible to make the operator polish and corrode the sample again, and it is possible to suppress execution of the quality evaluation process 5e using the cross section 102a in which the corroded state and polished state are not good.

In the polishing state determination process 5d (step S27 in FIGS. 1 and 10), the processor 5 of the present embodiment processes information indicating "accepted" or information indicating "failed" according to the aspect ratio of the object. output to That is, in the polishing state determination processing 5d, the processor 5 determines the polishing state of the cross section 102a based on the aspect ratio, which is the value indicating the flatness of the object included in the bright-side image generated by the image generation processing 5c. to output
If polishing marks are present in the cross section 102a, the polishing marks appear as objects in the light-side image. Since the polishing marks are linear, the aspect ratio of the object due to the polishing marks is large compared to the aspect ratio of the tissue grains.
If there are many objects with polishing marks among the objects included in the bright-side image generated by the image generation processing 5c (FIG. 1), the aspect ratio of the objects becomes relatively large. In other words, the aspect ratio of the object indicates the amount of polishing marks.
According to the above configuration, the judgment result regarding the polished state of the cross section 102a is output based on the aspect ratio of the object affected by the presence or absence of polishing marks during polishing, so that the polished state of the cross section 102a can be quantitatively judged. can be done. As a result, the polished state of the cross section 102a can be appropriately determined.

Further, in the present embodiment, the processor 5 determines the degree of corrosion of the cross section 102a based on the mode value (determined value) obtained from the histogram of the luminance value representing the brightness of each pixel constituting the target image and the appearance frequency thereof. The corrosion state determination process 5b for determining the state is executed, and if the determination result of the corrosion state determination process 5b is not "fail" (if the determination result of the corrosion state determination process 5b is "pass"), the image generation process 5c and polishing state determination processing 5d are executed. Accordingly, after quantitatively evaluating the corroded state of the cross section 102a, the polished state of the cross section 102a can be quantitatively evaluated.

Further, according to the above configuration, when the determination results of the corrosion state determination process 5b and the polishing state determination process 5d are not both "failed" (when they are "passed"), the processor 5 executes the quality evaluation process 5e. , an area image is acquired from the microscope image, and a bright side image in which the grain of the tissue in the cross section 102a is shown as a bright object and the other is shown as a dark area from the grayscale image based on the area image, and a grain of the tissue in the cross section 102a and a dark-side image showing a dark object and others showing a bright area, and based on the bright-side image and the dark-side image, an estimated physical property value of the cut piece 102 is obtained.

According to the above configuration, the corroded state and the polished state of the cross section 102a can be quantitatively determined, so the corroded state and the polished state of the cross section 102a can be appropriately determined. As a result, for example, a microscope image in which the cross section 102a is in a poorly corroded or polished state can be prevented from being used in the quality evaluation process, thereby suppressing a decrease in accuracy in estimating physical property values.

[About modification]
FIG. 16 is a diagram showing part of a flowchart of determination processing according to a modification.
In the corrosion state determination process 5b (FIGS. 1 and 10, step S23), the processor 5 of this modification obtains a determination value from a histogram based on luminance values representing the brightness of each pixel constituting the target image. This differs from the above embodiment in that the standard deviation obtained from the histogram is used in addition to the mode.
Note that FIG. 16 shows only steps S22 and S23, which are steps different from those in FIG. Other steps S21, S24 to S28 are the same as in FIG.

In this modification, the processor 5 obtains the standard deviation together with the mode of the histogram in step S22 in FIG.
Next, in the corrosion state determination process 5b, the processor 5 determines whether the mode is equal to or greater than the lower limit Th2 and equal to or less than the upper limit Th1, and whether the standard deviation is equal to or greater than the lower limit Th4 (step S23 in FIG. 16). ).
If the mode is not equal to or greater than the lower limit Th2 and equal to or less than the upper limit Th1, or if the standard deviation is not equal to or greater than the lower limit Th4, the processor 5 proceeds to step S25 (FIG. 10) and Information indicating "passed" is output to the worker, and the determination process is finished.

On the other hand, when the mode value is equal to or greater than the lower limit value Th2 and equal to or less than the upper limit value Th1, and the standard deviation is equal to or greater than the lower limit value Th4, the processor 5 proceeds to step S24 (FIG. 10) to continue processing.

The lower limit value Th4, which is the standard deviation threshold value set for the standard deviation, is set according to the material of the member to be evaluated, like the mode value. For example, the lower limit value Th4 is set to 15 in the case of the material (low-carbon alloy steel or low-carbon steel) of the member to be evaluated in this embodiment.
The lower limit Th4 is set as follows.

That is, the standard deviation of each of the histograms H1, H3, H5, H7, and H10 in FIG. Set the value Th4.
The standard deviation of histogram H1 in FIG. 12 is 10.56, the standard deviation of histogram H3 is 15.27, the standard deviation of histogram H5 is 20.39, the standard deviation of histogram H7 is 24.19, and the standard deviation of histogram H10 is 15. .45.
Based on this result, the lower limit Th4 for the standard deviation is set to 15.

The smaller the standard deviation value, the narrower the histogram peak width. Therefore, it can be said that the smaller the standard deviation value, the smaller the variation in brightness gradation included in the target image, and the smaller the amount of data represented by the gradation. For this reason, by setting a lower limit for the standard deviation, it is possible to relatively increase the variation in the brightness contained in the target image, and it is possible to secure a certain amount of data represented by gradation. The polishing state of the cross section 102a by the state determination processing 5d can be performed more appropriately.

In addition, in the present embodiment, the determination values include the mode and standard deviation of the histogram, and by using two determination values from different viewpoints, the corrosion state of the cross section 102a can be determined more appropriately. be able to.

In the above-described embodiment, only the mode value obtained from the histogram based on the brightness value representing the brightness of each pixel constituting the target image is used as the determination value. and the case of using the standard deviation obtained from the histogram, but it is also possible to use only the standard deviation as the judgment value.

〔others〕
The embodiments disclosed this time are illustrative in all respects and are not restrictive.
For example, the processor 5 of the above embodiment uses the aspect ratio as the value indicating the flatness of the object included in the bright side image in the polishing state determination process 5d (step S27 in FIGS. 1 and 10). As an example, the circularity of the object could be used instead of the aspect ratio, or the long axis of the object could be used. Further, a plurality of values selected from these values may be used in the polishing state determination process 5d as values indicating the flatness of the object.
It should be noted that the long axis of the polishing mark object is much larger than the long axis of the tissue grain object. Therefore, it is possible to determine the presence of polishing marks by the long axis of the object, which is a value indicating the flatness of the object.

In the above embodiment, the processor 5 of the quality evaluation device 4 executes the acquisition process 5a, the corrosion state determination process 5b, the image generation process 5c, the polishing state determination process 5d, and the quality evaluation process 5e. By causing the processor 5 to execute each process other than the quality evaluation process 5e, the quality evaluation device 4 can be used as a judgment device for judging the corrosion state and the polishing state.
The quality evaluation device 4 can also be used as a polishing state determination device by causing the processor 5 to execute only the acquisition processing 5a, the image generation processing 5c, and the polishing state determination processing 5d.
In this case, the processor 5 outputs information indicating "passed" or "failed" to the operator in the polishing state determination process 5d based on the aspect ratio of the object.

The scope of rights of the present invention is not limited to the above-described embodiments, but includes all modifications within the scope of equivalents to the configurations described in the claims.

1 Quality evaluation system 2 Microscope 3 Camera 4 Quality evaluation device 5 Processor 5a Acquisition process 5b Corrosion state determination process 5c Image generation process 5d Polished state determination process 5e Quality evaluation process 6 Storage device 6a Microscopic image 6b Threshold data 6c Quality evaluation model 7 Input Output unit 100 Raceway ring 102 Cut piece 102a Cross section 102a1 Raceway surface 106 Sample 106a Observation surface 255 Brightness G1 Target image G2 Target image G3 Target image G4 Target image G5 Target image H1 Histogram H10 Histogram H3 Histogram H5 Histogram H7 Histogram L Curve P1 Group of black circles P10 Black circle group P11 Black circle group P12 Black circle group P3 Black circle group P5 Black circle group P7 Black circle group

Claims (5)

Acquisition processing for acquiring a target image to be determined from a microscopic image of a metal surface that has been corroded by a etchant for structure observation after polishing;
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. an image generation process that generates at least one of the dark side images shown as regions;
polishing state determination processing for outputting a determination result regarding the polishing state of the metal surface based on the value indicating the flatness of the object included in the image generated by the image generation processing;
A polishing state determination device comprising a processing unit that executes 2. The polished state determination device according to claim 1, wherein the value indicating the flatness of the object includes at least one of the aspect ratio, circularity, and length of the long axis of the object. The processing unit determines a corrosion state of the metal surface, prior to the image generation processing, based on a histogram based on a luminance value representing brightness of each pixel constituting the target image and its appearance frequency. Execute the judgment process,
3. The image generation process according to claim 1, wherein at least one of the bright side image and the dark side image is generated when a determination result of the corrosion state determination process satisfies a predetermined condition. polishing state judgment device. A polished state determination method for determining the polished state of a metal surface corroded by a corrosive liquid for structure observation after polishing, comprising:
Acquiring a target image to be determined from the microscope image of the metal surface,
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. generating at least one of the dark side images shown as regions;
A polishing state determination method for outputting a determination result regarding a polishing state of a metal surface based on a value indicating the flatness of the object included in the generated image. A quality evaluation device for evaluating the quality of the metal based on a microscope image of the metal surface corroded by the etchant for structure observation after polishing,
Acquisition processing for acquiring a target image from the microscope image, which is an object for determining the polishing state of the metal surface;
From a grayscale image based on the target image, a light-side image in which the metal texture grains are shown as bright objects and others as dark areas, or a light side image in which the metal texture grains are shown as dark objects and others are bright. an image generation process that generates at least one of the dark side images shown as regions;
polishing state determination processing for outputting a determination result regarding the polishing state of the metal surface based on the value indicating the flatness of the object included in the image generated by the image generation processing;
When the judgment result of the polishing state judgment processing satisfies a predetermined condition, an area image of an area to be subjected to quality evaluation is obtained from the microscope image, and the light side image and the bright side image and the light side image are obtained from a grayscale image based on the area image. a quality evaluation process of generating a dark side image and obtaining physical property values of the metal based on the bright side image and the dark side image;
A quality evaluation device comprising a processing unit that executes

Images