JPH10274650A - Method and device for inspecting non-metal inclusion in metal material - Google Patents

Abstract

PROBLEM TO BE SOLVED: To determine the presence of a non-metal inclusion and its kind by selecting and scanning a visual field where it has been judged that a foreign object exists by a rough inspection and processing an image and then performing a neurofuzzy reasoning from the shape and color parameters of the foreign object. SOLUTION: The position of a sample stage 2 is adjusted and moved only for a visual field where a foreign object is recognized by a rough inspection, so that it is selected and scanned. An image that is picked up by a CCD camera 6 is processed by an image board 8 according to the instruction of a host personal computer 10. Based on the result, the host personal computer 10 judges whether a foreign object is a non-metal inclusion or not. When the object is a non-metal inclusion, its kind is discriminated and, for example, the number of the foreign objects, sizes, and area rates are obtained for each kind. The discrimination is performed by a neurofuzzy reasoning and the degree of the circle and the degree of the complication of the foreign object, the shape parameter of an aspect ratio, a color parameter of a histogram of three primary colors of red, green, and blue, the degree of dispersion of the histogram, and the degree of a pattern matching are adopted as the parameters, thus reducing an erroneous recognition and allowing an automatic inspection.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for automatically inspecting nonmetallic inclusions in a metal material such as steel.

[0002]

2. Description of the Related Art Since non-metallic inclusions in steel have a great effect on various material properties such as fatigue strength, an operator observes a mirror-polished sample with a metallographic microscope to determine the type of non-metallic inclusions. And required inspection data was calculated. As inspection data, cleanliness and the like defined in JIS G0555 are generally adopted. The degree of cleanliness is calculated by inserting a glass plate having 20 grid lines in each of the vertical and horizontal directions into the eyepiece, repeatedly inspecting the inspected surface at random, and counting the number of grid point centers occupied by inclusions. Things.

[0003] Observation of non-metallic inclusions with a metallographic microscope requires a high degree of judgment to determine the type of inclusions.
As the IS standard stipulates that 60 visual fields are measured in principle, patience is required because the number of visual fields is large. Therefore, a mental and physical load is imposed on the operator. In recent years, attempts have been made to automate inspection using image processing of a microscope image.

[0004]

In the conventional automatic inspection of non-metallic inclusions, the processing speed is slow and a long time is required because focusing is performed for each observation field of the metal microscope.
In addition, in image processing, the probability of erroneous recognition is high when discriminating non-metallic inclusions from attached matter such as dust and the type of non-metallic inclusions, and the reliability of inspection data is insufficient.

The present invention relates to a method and an apparatus for automatically inspecting non-metallic inclusions in a metal material such as steel, which greatly reduces the processing time as compared with conventional methods, and reduces non-metallic inclusions and dust. An object of the present invention is to reduce the probability of misrecognition and improve the reliability in discriminating the presence of a non-metallic inclusion and the type of nonmetallic inclusions, and make it possible to substitute for manual inspection.

[0006]

According to the present invention, there is provided a method for automatically inspecting a non-metallic inclusion by observing a mirror-polished sample surface of a metal material with a metallographic microscope. Performing alignment, rough inspection and precision inspection, and in focusing, the sample to be inspected is fixed on a sample stage whose position can be adjusted in the X direction and the Y direction and the Z direction in the height direction within the sample surface, and the sample surface is adjusted. Obtain the X-direction tilt and the Y-direction tilt, calculate the focal position of each visual field of the inspection range from the respective tilts, and in the rough inspection, adjust the sample position while adjusting the focal position of the calculation result for all the visual fields of the inspection range. Scan the entire field of view of the stage, process the image of the microscope image captured by the CCD camera and determine the presence or absence of foreign matter,
In the fine inspection, selective scanning is performed for each visual field determined to have foreign matter in the coarse inspection, and a shape parameter consisting of circularity, complexity and aspect ratio of the foreign matter, and a color consisting of a histogram of three primary colors of red, green and blue are obtained by image processing. Neuro-fuzzy inference using the parameters, the degree of dispersion of the histogram, and the degree of pattern matching as parameters is performed, and whether or not the foreign matter is a non-metallic inclusion, and if it is a non-metallic inclusion, the type thereof is determined. A method for inspecting nonmetallic inclusions in a metal material, characterized by outputting data.

A device according to the present invention for achieving the above object is a device for automatically inspecting a non-metallic inclusion by observing a mirror-polished sample surface of a metal material with a metallographic microscope. A sample stage controller for adjusting the position of the sample stage in the X direction and the Y direction within the sample surface, a Z-axis controller for adjusting the position of the sample stage in the Z direction of the height direction, and capturing a microscope image of the sample surface. CCD camera and arithmetic and control processing device, the arithmetic and control processing device has a focusing mechanism, a coarse inspection mechanism and a fine inspection mechanism, and the focusing mechanism is configured to detect the X-direction tilt and the Y-direction tilt of the sample surface. The inspection position is a mechanism that calculates the focal position of each field of view and operates the sample stage controller and the Z-axis controller. For viewing, actuates said sample stage controller and Z-axis controller, performs full field scanning of the sample stage, the CC
A mechanism for determining the presence or absence of foreign matter by performing image processing on the result of capturing the microscope image by the D camera, and the precision inspection mechanism operates the sample stage controller and the Z-axis controller, and determines that foreign matter is present in the coarse inspection. By performing selective scanning on the field of view, image processing is used to determine the circularity, complexity, and aspect ratio of the foreign matter, the color parameters composed of histograms of the three primary colors of red, green, and blue, and the variance and pattern matching of the histogram as parameters. A metallic material characterized by performing a neuro-fuzzy inference to determine whether or not the foreign matter is a non-metallic inclusion and, if it is a non-metallic inclusion, determine the type thereof and output necessary data It is a non-metallic inclusion inspection device inside.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The method of the present invention will be described with reference to a specific example shown in the block diagram of FIG. A sample stage 2, a Z-axis controller 4, and a CCD camera 6 are connected to the metal microscope 1. The sample stage 2 has a sample stage controller 3, and the CCD camera 6 has a camera power supply 7.
, Respectively.

The sample stage controller 3 and the Z-axis controller 4 are connected to a host personal computer 10 via a communication interface 5, and the CCD camera 6 and the camera power supply 7 are connected to the host personal computer 1 via an image processing board 8.
Connected to 0. The image processing board 8 is provided with an image monitor 9. Also, the host PC 10
A keyboard 11, a mouse 12, a PC monitor 13 and a printer 14 are connected to each other.

The position of the sample stage 2 is adjusted in the X and Y directions by the sample stage controller 3, and the position is adjusted in the Z direction by the Z-axis controller 4. The X direction and the Y direction are directions orthogonal to each other in the sample surface, and the Z direction is the height direction of the sample surface. The Z-axis controller 4 is obtained by adding a function of adjusting the position of the sample stage 2 in the Z direction to a controller of the main body of the metal microscope 1 such as lens switching, and controls the sample stage 2 via the metal microscope 1.

In order to carry out the method of the present invention, first, a mirror-polished metal material to be inspected is fixed to a sample stage 2 of a metal microscope 1. Next, as an initial input, a range to be inspected on the sample surface and inspection data to be output are input to the host personal computer 10 by the keyboard 11, and an initial inspection position on the sample surface is designated to the sample stage controller 3. Thereafter, an automatic inspection is performed according to the following procedure according to the method of the present invention.

1. Focusing 1-1) The sample stage controller 3 and the Z-axis controller 4 operate to adjust the position of the sample stage 2 in the XYZ directions, and focus on three visual fields among the four visual fields of the square or rectangular corner portion of the inspection range. , X at the center of each of the three visual fields
The YZ coordinate points are read, and the host PC 10 calculates the X-direction tilt and the Y-direction tilt of the inspection range. Here, X and Y are two orthogonal directions in the sample plane, and Z is a height direction. 1-2) XYZ coordinate points of the above three points, inclination in X direction, and Y
The Z coordinate point of the focal point is calculated for each visual field in the inspection range from the direction inclination.

2. Rough inspection 2-1) The sample stage 2 moves while adjusting the position of the entire visual field in the inspection range, and performs the full visual field scanning using the Z coordinate point of each visual field as the focal height. 2-2) The image processing result of the whole field of view taken by the CCD camera 6 is processed by the image processing board 8 according to the instruction of the host personal computer 10, and the presence or absence of foreign matter is determined by the host personal computer 10 from the result. The visual field position is stored.

3. Precise inspection 3-1) Sample stage 2 only for the above visual field with foreign matter
Is adjusted to perform selective scanning. 3-2) The imaging result of the field of view with the foreign matter by the CCD camera 6 is subjected to image processing by the image processing board 8 in accordance with the instruction of the host personal computer 10, and based on the result, the foreign matter is a non-metallic inclusion by the host personal computer 10. Determine whether there is
If it is a nonmetallic inclusion, its type is determined, and the number, size, area ratio, etc., of each type are measured. When the cleanliness specified in JIS G 0555 is output, the non-metallic inclusions occupy the grid points formed by 20 grid lines each in the vertical and horizontal directions of the glass plate inserted into the eyepiece of the metal microscope 1. Measure the number of grid points. 3-3) Calculate and display and record inspection data.

In the above procedure, the inspection data includes
In addition to the cleanliness of nonmetallic inclusions specified in JIS, the number of nonmetallic inclusions by type and size can be output.

Here, how to determine the X-direction tilt and Y-direction tilt of the sample surface in focusing and how to obtain the Z coordinate point of the focal point for each field of view will be described with reference to FIGS. It is assumed that the microscope field of view is divided as shown in FIG. 2 within the square or rectangle of the inspection range. The total number of visual fields in the X direction is Xmax, the total number of visual fields in the Y direction is Ymax, the number of visual fields whose observation visual field has moved in the X direction due to the movement of the sample stage 2 is i, and the number of visual fields that have moved in the Y direction is j.

When focusing on four visual fields at the corners of the inspection range, ie, three visual fields among A, B, C, and D, for example, A, B, and C, and setting the Z coordinate of the visual field of A to 0, the inclination in the X direction is shown in FIG. 3
.DELTA.AB / Xmax, and the inclination in the Y direction is .DELTA.
AC / Ymax. Since the inspection range can be assumed to be a plane, the Z coordinate point of the focal position in an arbitrary observation field of XY coordinates (i, j) is Z = (ΔAB / Xmax) · i + (ΔAC / Ymax) · j ( 1) can be expressed as The focusing of each of the fields A, B, and C can be performed by an autofocus logic generally used in a conventional microscope.

In the determination of the presence or absence of foreign matter in the rough inspection, the image pickup result of the CCD camera 6 is binarized by image processing. If the area ratio of the foreign matter to the visual field exceeds a predetermined threshold value, it is determined that foreign matter is present. The time required for one visual field is several hundred msec.

In the detailed inspection, it is determined whether or not the foreign substance is a non-metallic inclusion, and if the foreign substance is a non-metallic inclusion, the type of the inclusion is determined by performing image processing on the imaging result of the CCD camera 6 and performing neuro- Performed by fuzzy inference. Neuro
The fuzzy parameters include a shape parameter including the circularity, complexity, and aspect ratio of the foreign matter, and red, green, and blue.
A color parameter including a histogram of primary colors, a degree of dispersion of the histogram, and a degree of pattern matching are employed.

The type of nonmetallic inclusion is JIS in the case of steel.
A system, B system and C system defined in G 0555, and if necessary, further A1 system, A2 system, B1 system, B2 system, C1 system,
It can be classified as C2. As the foreign matter other than the nonmetallic inclusions, it is possible to determine a convex portion such as attached dust and a concave portion such as a hole or a flaw. By the way, A-type inclusions are those that are viscously deformed by processing (sulfides, silicates, etc.), and if necessary, are further divided into A1 (sulfides) and A2 (silicates). ing.
The B-based inclusion is a group of discrete inclusions (such as alumina) which are arranged in a group in the processing direction and are discontinuous.
In the case of steel containing one or more kinds of Zr, if necessary, B1 type (oxide type such as alumina)
And B2 (carbonitride of Nb, Ti, Zr). C-based inclusions are irregularly dispersed without viscous deformation (eg, particulate oxides).
In steels containing one or more of Ti and Zr,
If necessary, it is further specified that the material is further divided into a C1 type (oxide type) and a C2 type (Nb, Ti, Zr carbonitride type).

The circularity a, complexity b and aspect ratio c of the shape parameters are as follows: a = 4π × (area) / (perimeter) ² b = (perimeter) ² / (area) c = (major axis) / ( Minor axis).

The degree of dispersion of the histogram is as shown in FIG.
It is defined by the ratio of the area in the set range indicated by oblique lines to the entire area of the histo pattern. In addition, the pattern matching degree of the histogram is obtained by preparing a standard pattern of the histogram for various known non-metallic inclusions in advance, and calculating the ratio of the area difference when the histogram obtained for the unknown foreign substance is superimposed to the area of the standard pattern. Is defined by

FIG. 6B shows an example of the pattern matching degree.
As described above, the area difference between the solid standard pattern and the histogram pattern of the unknown foreign substance indicated by the broken line, that is, the ratio between the area of the hatched portion and the area of the standard pattern is the pattern matching degree. When the peak position of the histogram pattern of the unknown foreign substance does not coincide with the peak position of the standard pattern as indicated by the broken line in FIG. And determine the area difference.

The neuro-fuzzy inference in the method of the present invention can be performed by a method generally used conventionally, employing the above parameters. That is, for each of the above parameters, the measured values of the known foreign matter are stored in advance, and the measured values of the unknown foreign matter are sequentially compared and discriminated. As for the determination result, the skilled operator determines whether the recognition is correct or incorrect, and repeatedly learns, thereby increasing the correct recognition rate and improving the determination speed.

Next, as shown in the example of FIG. 1, the apparatus of the present invention comprises a metal microscope 1, a sample stage 2, a sample stage controller 3, a Z-axis controller 4, a CCD camera 6, and an arithmetic control processor. The arithmetic and control unit is
It has a focusing mechanism, a coarse inspection mechanism and a precision inspection mechanism.

The focusing mechanism is a mechanism for calculating the focal position of each visual field in the inspection range from the tilt of the sample surface in the X and Y directions and operating the sample stage controller 3 and the Z-axis controller 4 in FIG. In the example, a host personal computer 10, a communication interface 5, a sample stage controller 3, a Z-axis controller 4, an image processing board 8,
It comprises a camera power supply 7 and a CCD camera 6.

The coarse inspection mechanism operates the sample stage controller 3 and the Z-axis controller 4 for the entire visual field of the inspected area, scans the entire visual field of the sample stage 2, and obtains the microscopic image captured by the CCD camera 6. A mechanism that determines the presence or absence of foreign matter by performing image processing.
In the example of FIG. 1, the host computer 10 includes a host computer 10, a communication interface 5, a sample stage controller 3, a Z-axis controller 4, an image processing board 8, a camera power supply 7, and a CCD camera 6.

The precision inspection mechanism operates the sample stage controller 3 and the Z-axis controller 4 to perform selective scanning for each visual field determined to have foreign matter in the coarse inspection, and to perform image processing to determine the circularity, complexity and aspect of the foreign matter. A shape parameter consisting of a ratio and a color parameter consisting of a histogram of the three primary colors of red, green, and blue, and performing a neuro-fuzzy inference using the variance and the pattern matching degree of the histogram as parameters, to determine whether or not the foreign matter is a nonmetallic inclusion, If it is a non-metallic inclusion, determine its type,
In the example of FIG. 1 which is a mechanism for outputting necessary data, the precision inspection mechanism includes a host personal computer 10, a communication interface 5, a sample stage controller 3, a Z-axis controller 4, an image processing board 8, a camera power supply 7, a CCD camera 6, It comprises a PC monitor 13 and a printer 14. The operation of each mechanism is as described in the method of the present invention.

[0029]

EXAMPLE Forty samples of rolled bearing steel were cut out, and the surface parallel to the rolling direction was mirror-polished, and nonmetallic inclusions were inspected by the method and apparatus of the present invention having the structure shown in FIG. The area to be inspected was a square of 5 mm x 5 mm for each sample, and the microscope was examined at 220 magnifications at a magnification of 400 times. The inspection time was 4 minutes per sample. For the largest foreign substance present in each sample, the result of automatic discrimination by the method and apparatus of the present invention and the result of discrimination by a skilled operator are compared, and if the discrimination of a skilled operator is correct, as shown in Table 1, the positive recognition number Is 40 in total
The number was 36 out of the 90%, and the correct answer rate was 90%.

FIGS. 7, 8 and 9 show examples of determination. In each figure, the left side is an image binarized by image processing after imaging with a CCD camera, and the right side is a histogram of three primary colors. FIG. 7 discriminates sulfide (sulfide), FIG. 8 discriminates oxide (oxide), and FIG. 9 discriminates adhered dust.

[0031]

[Table 1]

[0032]

By employing the method and the apparatus of the present invention, it becomes possible to automate the inspection of non-metallic inclusions, which has conventionally imposed a great mental and physical load on the operator, at a practical level. That is, the conventional automation process requires a long time, and the discrimination result of the type of the inclusion has a high probability of erroneous recognition and is insufficient in reliability. It is possible.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration example of a method and an apparatus of the present invention.

FIG. 2 is an explanatory view of a focusing mechanism in the method and apparatus of the present invention.

FIG. 3 is an explanatory diagram of an X-direction tilt of a sample surface in the method and the apparatus of the present invention.

FIG. 4 is an explanatory view of the Y-direction tilt of the sample surface in the method and the apparatus of the present invention.

FIG. 5 is an explanatory diagram of a histogram dispersion degree in the method and apparatus of the present invention.

FIG. 6 is an explanatory diagram of a pattern matching degree of a histogram in the method and apparatus of the present invention.

FIG. 7 is an image and a histogram showing an example of a determination result in the embodiment of the method and the apparatus of the present invention.

FIG. 8 is an image and a histogram showing another example of the determination result in the embodiment of the method and the apparatus of the present invention.

FIG. 9 is an image and a histogram showing another example of the determination result in the embodiment of the method and the apparatus of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Metal microscope 2 ... Sample stage 3 ... Sample stage controller 4 ... Z-axis controller 5 ... Communication interface 6 ... CCD camera 7 ... Camera power supply 8 ... Image processing board 9 ... Image monitor 10 ... Host personal computer 11 ... Keyboard 12 ... Mouse 13 Monitor for PC 14 Printer

Claims (2)

[Claims] 1. A method for automatically inspecting non-metallic inclusions by observing a mirror-polished sample surface of a metal material with a metallographic microscope, performing focusing, coarse inspection and precision inspection. The sample to be inspected is fixed on a sample stage that can be adjusted in position in the X direction and the Y direction in the plane and the Z direction in the height direction, and the X direction inclination and the Y direction inclination of the sample surface are obtained. The focus position of each field of view is calculated, and in the coarse inspection, the entire field of view of the inspection target area is scanned in the entire field of view of the sample stage while being adjusted to the focus position of the calculation result. Image processing to determine the presence or absence of foreign matter,
In the fine inspection, selective scanning is performed for each visual field determined to have foreign matter in the coarse inspection, and a shape parameter consisting of circularity, complexity and aspect ratio of the foreign matter, and a color consisting of a histogram of three primary colors of red, green and blue are obtained by image processing. Neuro-fuzzy inference using the parameters, the degree of dispersion of the histogram, and the degree of pattern matching as parameters is performed, and whether or not the foreign matter is a non-metallic inclusion, and if it is a non-metallic inclusion, the type thereof is determined. A method for inspecting nonmetallic inclusions in a metal material, comprising outputting data. 2. An apparatus for automatically inspecting non-metallic inclusions by observing a mirror-polished sample surface of a metal material with a metal microscope, wherein the metal microscope, the sample stage of the microscope are moved in the X direction and Y direction in the sample surface. Stage controller for adjusting the position in the Z direction, a Z-axis controller for adjusting the position of the sample stage in the Z direction in the height direction, a CCD camera for capturing a microscope image of the sample surface, and an arithmetic and control unit The arithmetic and control unit is a focusing mechanism,
It has a coarse inspection mechanism and a fine inspection mechanism, and the focusing mechanism calculates the focal position of each visual field in the inspection range from the X-direction tilt and the Y-direction tilt of the sample surface, and operates the sample stage controller and the Z-axis controller. The coarse inspection mechanism operates the sample stage controller and the Z-axis controller for the entire visual field of the inspection range, performs full-field scanning of the sample stage, and images the microscope image captured by the CCD camera. The precision inspection mechanism operates the sample stage controller and the Z-axis controller, performs selective scanning for each field of view determined to have foreign matter in the coarse inspection, and performs image processing. Shape parameters consisting of the circularity, complexity and aspect ratio of the foreign matter, and Neuro-fuzzy inference is performed using color parameters consisting of histograms of the three primary colors of red, green and blue, the degree of dispersion and the degree of pattern matching of the histogram as parameters, and whether or not the foreign substance is a non-metallic inclusion is determined. In this case, a non-metallic inclusion inspection apparatus in a metal material is characterized by a mechanism for determining the type and outputting necessary data.