JP3071956B2 - Vickers hardness automatic reading device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a Vickers hardness tester, and more particularly to a method for indenting a surface of a test piece (sample) to form a depression, and measuring the hardness from the surface area of the depression and the indentation load of the indenter. In the Vickers hardness tester, the surface area is automatically measured, and the Vickers hardness can be automatically measured based on the measurement.
The present invention relates to an automatic Vickers hardness reading device.

[0002]

2. Description of the Related Art Conventionally, in the measurement of Vickers hardness,
The measurement of the surface area of the depression formed on the sample surface by the indentation is performed by measuring the diagonal length of the depression with the naked eye using a measuring microscope. By the way, the measurement with the naked eye has the disadvantage that there are individual differences in measurement, the measurement time is long, and errors occur due to the degree of fatigue of the measurer. However, when a person performs measurement, there is an advantage that the dent can be selectively recognized from scratches and dirt other than the dent.

In order to solve the above-mentioned drawbacks, there has been proposed an apparatus for automatically measuring the hardness by mounting a TV camera on an eyepiece of a hardness meter to project a dent, performing an appropriate image processing on the image, and performing an appropriate image processing. No. 4-59424). However, this device has a problem that the position of the depression on the image needs to be substantially at the center of the screen, and when there is a stain of a certain size or more, the top of the depression is easily misidentified. Furthermore, in the case of the above-mentioned Japanese Patent Application No. 4-59424, when only the measurement of the surface area of the hollow is automated, the position of the hollow is moved to the center of the screen by the measurer, and a judgment is made as to whether the measurement is a false recognition measurement or a steady measurement. Although the first problem described above cannot be solved, when used as a fully automatic measuring device that automatically performs from the indentation of a number of points to the hardness measurement, the reliability of the measurement results is inferior. Lack of sex.

[0004]

The present invention seeks to overcome this shortcoming of the prior art. That is, the present invention uses a TV camera attached to the eyepiece of a hardness meter to project a dent formed on the surface of a sample, digitize the image of the dent into multi-valued gradations, and apply a multi-valued digital image to the dent. An optimum threshold value is detected, a binary image is obtained using the threshold value, and a two-dimensional noise removal process is performed on the binary image to selectively reduce the influence of scratches and dirt. The approximate positions in the screen of the four sides are detected, the indentation boundary point is detected from the approximate positions, the boundary point is regressed to four polynomial curves, and the intersection of these four polynomial curves is defined as the indentation vertex. Estimate, calculate the diagonal area by calculating the length of two diagonal lines that intersect each other from the coordinates on the four vertices, calculate the Vickers hardness from the dimple area and the pressing load acting on the indenter. Indentation in measurement A Vickers hardness automatic reading device that has a reliability that can withstand use in fully automatic measuring devices while resolving the problem of reading errors due to scratches and dirt etc. under recognition capabilities close to the ability The purpose is to provide.

[0005]

In order to achieve the above object, the automatic Vickers hardness reading apparatus of the present invention is capable of forming a Vickers depression by pushing a quadrangular pyramid-shaped indenter into the surface of a sample under a predetermined load. And a TV camera attached to the eyepiece of the optical microscope provided by the tester,
An A / D converter for analog-to-digital conversion of a video signal on the surface of the sample including the depressions projected by the TV camera in multi-valued gradation; An image storage unit that stores a value digital image, a threshold value detection unit that detects an optimal threshold value for the multivalued digital image, and the multivalued digital image based on a threshold value from the threshold value detection unit. , A two-dimensional noise removing unit that removes points outside the dent boundary from the binary image obtained by the binary converting unit, and a two-dimensional noise removing unit that A depression position detection unit for detecting the center of the depression and the approximate positions of the four sides in the screen from the value image, and a boundary for detecting the depression boundary based on the four sides of the depression and the binary image detected by the depression position detection unit Detector and four boundaries detected by the boundary detector A vertex estimating unit that returns to the curve and estimates the position on the coordinates of the concave vertex from the intersection of each curve, and a hardness calculation that calculates the Vickers hardness from the position of the vertex estimated by the vertex estimating unit and the indentation load of the indenter. It is characterized by having a department.

[0006]

In the above-described automatic Vickers hardness reading apparatus of the present invention, the A / D converter converts the video signal from the TV camera into a multi-value digital image. In the binary conversion unit,
The multi-valued digital image compares the threshold value detected by the binarization threshold value detection unit with the luminance of the pixel of the multi-valued digital image of interest, and determines “1” (white) and “0” (black). Is converted into a binary image. In the two-dimensional noise removing unit, a pixel that is determined not to be on the four sides of the boundary of the focused pixel depression is removed as noise. The depression position detection unit determines the approximate position of the depression by obtaining the center of a virtual square represented by four straight lines passing through the four sides of the depression from the noise-removed binary image, and further determines the boundary detection unit. Then, the indentation boundary is detected based on the four sides of the indentation detected by the indentation position detection unit and the binary image.

Further, the vertex estimating section regresses the position of the boundary pixel detected by the concave boundary detecting section to a higher-order curve for each side by a least square method, and calculates the intersection of the regression line from
The positions on the coordinates of the four vertices of the depression are obtained. Then, the hardness detection unit calculates the length of two diagonal lines intersecting from the coordinates on the four vertices obtained by the vertex estimation unit, and calculates the Vickers hardness Hv based on the calculated value. Equation 1] is performed based on the equation.

Hv = 1.854 P / d ² (kgf / mm ² ) where P is the indenter load and d is the average length of two diagonal lines.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An automatic Vickers hardness reading device as one embodiment of the present invention will be described below with reference to the drawings.
FIG. 1 is a schematic plan view showing a hardness calculator as a block diagram, FIG. 2 is an image of a Vickers dent formed on the surface of the sample, and FIG. FIG. 4 is an image showing a state in which a pixel of interest is detected starting from a cross position for detecting a depression position, and FIG. 5 is a diagram showing a state where M = 8 in a square area of 8 × 8 with respect to the state of FIG. 5 is an image from which noise has been selectively removed by performing two-dimensional noise removal, FIG. 6 is an image obtained by detecting a dent position to detect a virtual square, FIG. 7 is an image obtained by detecting a boundary, and FIG. 8 is a two-dimensional curve. The cross position is used as the vertex estimation in the image for which the vertex estimation has been performed. 9 is an image corresponding to FIG. 8 showing a state in which the depression position is greatly deviated from the center of the screen, FIG. 10 is an explanatory view of a measurement example of a diagonal length, and FIG. 6 is a graph showing test results obtained by repeating automatic reading twice.

As shown in FIG. 1, the automatic Vickers hardness reading apparatus according to this embodiment includes a Vickers hardness tester 1 and a hardness calculator 2 having a built-in image processing unit. The hardness tester 1 includes an XY automatic stage type sample table 12
And an indenter 11.
The surface of a test piece (sample) 10 placed on
(See FIG. 2), and a TV camera 13 is attached to the eyepiece. The hardness calculator 2 includes an A / D conversion unit 20 (the function and the like will be described later in detail, the same applies hereinafter), an image storage unit 21, a threshold value detection unit 23, a binary conversion unit 22, a two-dimensional noise removal. Part 2
4, a depression position detecting unit 25, a boundary detecting unit 26, a vertex estimating unit 27, a calculating unit 28, a hardness calculating unit 29, and a printer 30 are provided.

The above-mentioned A / D converter 20 disperses the video signal of the cavity 3 transmitted from the TV camera 13 via the video signal connection line 14 into, for example, 512 × 512 pixels, and separates each pixel into It has a function to convert to a multi-level digital image signal according to the brightness of the image. The A / D conversion unit 20 is provided in association with a threshold value detection unit for binarization, or an arithmetic unit.
To adapt the conventional technology (Japanese Patent Application No. 4-54924) to 28, 256
A resolution on the order of gradation is required. The A / D converted multi-level digital image signal is stored in the multi-level image storage unit 21 as 512 × 512 pixels. The multi-valued image storage unit 21 stores one
Each pixel has the same resolution as the A / D converter 20.

Next, the threshold detecting section 23 uses the following two technical means to detect an optimum binarized threshold. One is that when a luminance distribution histogram of a multi-valued digital image is divided into two groups by a certain threshold value, a binarization threshold value is set to a value that maximizes the variance between the two groups. Although this technical solution gives good results in most cases, it has the disadvantage of being time-consuming. Another technical means is to use a threshold value at which the number of white and black pixels when binarized (black and white) becomes a predetermined ratio as a binarization threshold value. Good results are obtained when the contrast of the image exceeds a certain level, and the time required for detection is short. Therefore, it is necessary for a person to judge and set the above-mentioned ratio depending on the relationship between the surface condition of the test piece, the magnification of the measuring microscope, the illumination illuminance, and the like. In the case of this embodiment, the better of these two technical means is selected.

The binary conversion unit 22 includes a binary threshold detection unit
The threshold value detected by step 23 is compared with the luminance signal of the pixel of the multi-valued digital image of interest. If the threshold value of the pixel signal is lower, “1” (white), otherwise “ 0 ”(black)
An operation of converting the signal into a signal is performed. FIG. 3 shows a binary image 4 that has been binarized using a threshold value that maximizes the variance between white / black groups. The binary image signal binarized by the binary conversion unit 22 is sent to the two-dimensional noise removal unit 24 next. 2
The following operation is performed in the dimensional noise removing unit 24.
In general, the shape of the depression is a substantially square, and the image has four boundaries rotated by 45 ° on the screen as shown in FIG. That is, the four sides of the boundary are displayed on the screen as straight line segments in the direction of approximately 45 ° or 135 ° with respect to the horizontal line or the vertical line. Therefore, the two-dimensional noise removing unit 24 counts the number of pixels where the value of 2N-1 pixels on the diagonal line of the N × N square area is “0” (black) around the pixel of interest on the binary image. If the number is not smaller than a predetermined number M, it is determined that the pixel of interest is not the four sides of the depression, thereby removing noise. Note that binarization or two-dimensional noise removal of a multi-valued digital image is performed by
The measurement may be performed on the entire screen, or may be performed partially or in units of one pixel of interest as needed.

FIG. 5 shows an 8 × 8 square area, M = 5, and two-dimensional noise in the state of FIG. 4 (a state in which a pixel of interest is detected starting from a cross position to detect a depression position). The image after removal has been shown, indicating that noise has been selectively removed. Then, the image from which the two-dimensional noise has been removed by the two-dimensional noise removing unit 24 is sent to the next depression position detecting unit 25. The following operation is performed in the depression position detection unit 25. That is, 2
The effect of the dimensional noise removing unit 24 is not perfect. For example
The scratches in the 45 ° direction cannot be removed, and the inside of the hollow, the scratches having a large size, and the inside of the dirt cannot be removed.

Therefore, on the binary image, starting from an arbitrary position (eg, the center) on the screen, the first "0" (black) changes from right "0" (black) to "1" (white) rightward. Attention is paid to the pixel, and if the pixel is not subjected to the two-dimensional noise removal processing, 2
If a two-dimensional noise removal process is performed, and if the noise is not removed, a virtual 2
The straight lines A and B of the book are registered (see FIG. 6). The same process is performed in the left direction, and two virtual straight lines C and D are registered. This process is performed for all pixel positions on the screen vertical line passing through the starting point. Then, of the straight lines registered corresponding to the right direction of 45 °, the straight line having the maximum number of times of overlapping registration is selected. Similarly, a straight line group corresponding to 135 ° rightward, 45 ° leftward, and 135 ° leftward is selected. In this process, straight line registration may be performed after all the pixels of interest are selected.

The same process is performed in the vertical direction for all pixel positions on the horizontal line of the screen passing through the starting point, and four straight lines are selected. Therefore, since there are two straight lines A to H selected from the left, right, up, and down directions respectively corresponding to the four sides of the depression, the combination that can represent the four sides of the depression is
There are 16 ways. Therefore, the combination closest to the square is selected (the ratio of the diagonal lengths is close to 1.0). Then, the center on the screen of the virtual square represented by the combination of the virtual straight lines is obtained, and if the center coincides with the starting point, it can be estimated that the virtual square gives the approximate size of the depression, and 4 sides 4A-4D with 4 hollows
Is given, and it is estimated that the center gives the approximate position of the depression. If they do not match, the above operation is repeated until the center is matched as a new starting point, thereby detecting the position of the depression. In the boundary detection unit 26,
The depression boundary is detected from the detected four sides 4A to 4D and the binary image by the following operation.

That is, the pixel "1" on the binary image at one end of the upper left line segment of the virtual square giving the approximate position of the upper left side of the depression
If it is (white), find the pixel that first becomes "0" (black) to the right of this pixel, and if it is "0" (black), find the pixel that first becomes "1" (white) to the left of this pixel locate. The pixel is subjected to two-dimensional noise processing to reduce the influence of noise.
By repeating this processing up to the other end of the line segment, the upper left boundary 4C of the hollow can be detected. Similarly, the upper right boundary 4B is detected from the upper right line segment of the virtual square, and is used as data for estimating the upper vertex 5 (see FIG. 7). The lower left boundary is concavely detected in the vertical direction from the upper left line of the virtual square, and the lower left boundary 4D is detected in the vertical direction from the lower left line of the virtual square, and is used as data for estimating the left vertex 6. By performing similar processing on the other two sides of the depression, the boundary 4A for estimating the depression vertices 7 and 8 can be detected.

Next, the vertex estimating unit 27 determines the positions of the boundary pixels on the four sides of the depression by a higher-order curve (eg, a quadratic curve) for each side.
Then, an operation of calculating the coordinates of the vertices of the depression by calculating the intersection of the regression line by performing regression by the least square method is performed. In the case of a higher-order curve, there may be a plurality of intersections, but the one closest to the vertex of the virtual square is adopted and estimated as the vertex. That is, usually, the depression is often distorted from the square. In particular, each side of the depression is not a straight line due to the material, hardness, etc. of the sample. On the other hand, in order to determine the correct Vickers hardness, it is necessary to accurately determine the positions of the four vertices (corners) of the depression. For this purpose, in the hollow image subjected to noise removal and binarization, an approximate expression of the four sides forming the outline of the hollow [this approximate expression is not a straight line with the four sides of the outline of the hollow, but a curve approximated to this. (For example, regression to a higher-order curve such as a quadratic curve) J to R
(See FIG. 8), and intersections 5 to 8 of these four equations
Is performed to calculate the coordinates of.

FIG. 9 shows a case where the image of the depression is shifted from the center of the screen. In the hardness calculator 29, first, the vertex estimator
From the positions on the coordinates of the four vertices 5 to 8 obtained in 27, the lengths of two diagonal lines intersecting each other are calculated. On the other hand, in parallel with this, the calculation unit 28 calculates the method described in Japanese Patent Application No. 4-59424 from the four sides (boundaries) 4A to 4D of the depression obtained by the depression position detection unit 25, that is, in FIG. Boundary detection is performed while the measurement axis 31 is set vertically and shifted little by little, and two upper and lower vertex positions are obtained from the two uppermost or lowermost points among the obtained boundary points.

Similarly, the measurement axis is set in the horizontal direction, and the boundary is detected while being shifted little by little, thereby obtaining the left and right vertex positions. Then, an operation for calculating the length of two intersecting diagonal lines from the positions of the four points obtained by the above means is performed. Further, the hardness calculator 29 calculates the ratio of the lengths of the diagonal lines calculated by the above two calculations, and selects the value whose value (the ratio of the lengths of the diagonal lines) is closer to 1. Then, from the lengths of the two selected diagonals,
An operation for calculating the Vickers hardness Hv is performed according to [Equation 2].

Hv = 1.854 P / d ² (kgf / mm ² ) where P is the indentation load of the indenter d is the average length of two diagonal lines Further, the calculation result is printed out from the printer 30. Thus, according to the device of this embodiment,
The correct Vickers hardness of the sample can be read automatically.

FIG. 11 shows a test result obtained by repeatedly performing automatic reading 500 times continuously on a sample having a Vickers hardness of 200 using the apparatus of this embodiment. When the graph of FIG. 11 was read, the following result was obtained as the average value of the Vickers hardness Hv: Hv (average) = 200.16 standard deviation = 1.44. From this, it can be seen that according to the present apparatus, high reliability and reproducibility and stable measurement can be performed.

[0021]

As described above, according to the automatic Vickers hardness reading apparatus of the present invention, the following effects and advantages can be obtained. (1) Because an optimal threshold value is detected for a multivalued digital image of a depression, a binary image is obtained using this value, and two-dimensional noise removal processing and depression recognition are performed on the binary image. In addition, the problem of the reading error of the dent due to the scratch on the sample surface can be solved. (2) The four sides of the hollow are regressed to four polynomial curves, the coordinates of the four vertices of the hollow are determined from the four intersections of these regression curves, and the hollows intersect each other from the positions of the vertices. Since the two diagonal lengths are calculated, the hardness can be measured even when the apex of the dent is not clear due to scratches or dirt. (3) Due to the above reasons (1) and (2), a highly reliable automatic Vickers hardness reader can be obtained.

[Brief description of the drawings]

FIG. 1 is a schematic plan view of an automatic Vickers hardness reading apparatus as one embodiment of the present invention, showing a hardness calculator as a block diagram.

FIG. 2 is a photograph instead of a drawing showing a Vickers dent formed on the surface of a sample.

FIG. 3 is a photograph in place of a binary drawing binarized using a threshold value that maximizes the variance between white / black groups.

FIG. 4 is a photograph instead of a drawing showing a state in which a pixel of interest is detected starting from a cross position for detecting a depression position.

5 is a square area of 8 × 8, M = 5 with respect to the state of FIG. 4;
2 is a photograph replacing a drawing from which two-dimensional noise has been removed.

FIG. 6 is a photograph instead of a drawing showing a state in which a dent position is detected and a virtual square is detected.

FIG. 7 is a photograph instead of a drawing showing a state where boundary detection has been performed.

FIG. 8 is a photograph instead of a drawing showing a state in which vertex estimation is performed using a quadratic curve.

FIG. 9 is a photograph instead of a drawing showing a state in which the depression position is greatly deviated from the center of the image, corresponding to FIG. 8;

FIG. 10 is an explanatory diagram of a measurement example of a diagonal length.

[Fig. 11] Continuous 5 for test piece with Vickers hardness 200
It is a graph which shows the test result which performed the automatic reading of 00 times repeatedly.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Hardness tester 2 Hardness calculator 3 Recess 4 Binary image 5, 6, 7, 8 Apex 10 Sample 11 Indenter 12 Sample stand 13 TV camera 14 Video signal connection line 20 A / D conversion unit 21 Image storage unit 22 Binary Conversion unit 23 Threshold detection unit 24 Two-dimensional noise removal unit 25 Indentation position detection unit 26 Boundary detection unit 27 Vertex estimation unit 28 Operation unit 29 Hardness operation unit 30 Printer 31 Measurement axis J-R Higher-order curve

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G01N 3/00-3/62

Claims (1)

(57) [Claims] 1. A tester capable of forming a Vickers cavity by pressing a quadrangular pyramid-shaped indenter into a surface of a sample with a predetermined load, a TV camera attached to an eyepiece of an optical microscope provided with the tester, An A / D conversion unit that performs analog-to-digital conversion of a video signal on the surface of the sample including the depression, which is projected by the TV camera, in multi-valued gradation;
An image storage unit that stores the multivalued digital image converted by the D conversion unit; a threshold value detection unit that detects an optimal threshold value for the multivalued digital image; A binary conversion unit for binarizing the multi-valued digital image based on a threshold value, a two-dimensional noise removal unit for removing a point outside a concave boundary from the binary image obtained by the binary conversion unit,
A depression position detection unit that detects the center of the depression in the screen and the approximate position of the four sides from the binary image obtained by the two-dimensional noise removal unit; and the four sides and the binary value of the depression detected by the depression position detection unit A boundary detection unit that detects a concave boundary with an image, a vertex estimating unit that regresses the boundary detected by the boundary detection unit into four curves, and estimates a position on the coordinates of the concave vertex from the intersection of each curve; An automatic Vickers hardness reading device, comprising: a hardness calculating unit that calculates Vickers hardness from the position of the vertex estimated by the vertex estimating unit and the pressing load of the indenter.