JPH09281055A - Inspection method for chip - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an inspection method by which a defective chip in an outline part can be detected by an image processing method with reference to an object to be inspected, which comprises a plurality of outline lines composed of different kinds of lines. SOLUTION: Outline coordinate data on an object to be detected is found by an image processing operation, an inspection vector is installed on the contour coordinate data, a prescribed vector out of it is used as a reference vector, an angle which is formed by the inspection vector with reference to the reference vector is computed, and the conversion point of the outline shape in the object to be inspected and its shape in the mean time are found. Then, an outline approximate line in which the outline shape is changed into a numerical formula on the basis of the data is found, the outline coordinate data is scanned on the basis of the outline approximate line, a chip which exists in the outline part of the object to be inspected is detected, and its size is computed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention image-processes a chip in a contour portion of an inspection object such as a cutting tool whose contour shape is formed by a combination of straight lines and curved lines. Regarding how to inspect.

[0002]

2. Description of the Related Art A method for inspecting a defect in a linear peripheral portion by image processing is disclosed in Japanese Patent Application Laid-Open No. 1-263774 (hereinafter referred to as Conventional Example 1). This is because the image pickup signal of the object to be inspected is stored in the image memory as binarized pixel data, and the coordinates of the binarized pixel data representing the peripheral portion are also stored in the scanning area set so as to surround the linear peripheral portion. In addition, after calculating a linear equation by the least-squares method to form a contour approximation line, pixels that are apart from the calculated contour approximation line by a predetermined distance or more are excluded from the pixel data coordinates, and the contour approximation line is calculated again. By repeating the above steps to obtain a highly accurate contour approximation line, calculate the intersections of the contour approximation lines that are adjacent to each other, and set multiple scanning lines at a predetermined pitch based on the intersection coordinates. Then, the intersection of the scan line and the binarized pixel data representing the peripheral portion is obtained, the notch is recognized from the difference between the adjacent intersection coordinates, and the width and depth of the notch are calculated and compared with the determination value. This is a method of inspecting for the presence or absence of a chip. As a method of identifying the shape of the contour portion, which is not a technique for directly detecting a chip, a line type of a figure is disclosed in Japanese Patent Laid-Open No. 1-1116772 (hereinafter referred to as Conventional Example 2) from image data of a linear figure imaging signal. There is disclosed a method of recognizing whether the is linear or circular. This reads the figure as binary image data,
Image processing is performed to obtain contour coordinate point data, and the obtained coordinate point data is divided into a plurality of short lines and represented by a short vector,
By calculating whether the displacement angle of the short vector with respect to the previous short vector is approximately zero, positive or negative, the line type is a straight line if it is approximately zero, a counterclockwise arc if positive,
If it is negative, it is a method to recognize it as a clockwise arc.

[0003]

The contour of the cutting edge of a cutting tool often has a complicated shape that cannot be displayed by one function, which is a combination of straight lines and circles or other curved lines. On the other hand, in Conventional Example 1, the contour is formed only by a straight line, and cannot be applied to a portion including another curve. In addition, this chipping detection method is a method of obtaining the difference in the scanning pitch line direction based on the intersection of the scanning pitch line drawn vertically or horizontally in the screen and the binarized pixel data representing the peripheral portion, If the target linear peripheral portion is inclined, there is a problem in that an error occurs between the calculated size of the chip and the actual size of the chip. In Conventional Example 2,
A method for discriminating whether the imaged line type is a linear line or a circular curve is described, but no reference is made to the defect detection. In addition, the identification method is affected by minute defects / chips and surface roughness existing in the peripheral part of the contour line of the actual inspection object, and by the quantization error of the pixel, and the difference should be made at the place where it should be a straight line. Often does not show substantially zero, or shows almost zero when it should be a curve, and thus there is a problem in the accuracy of line type identification. It is an object of the present invention to recognize the contour shape of an object having a complicated contour shape, which is difficult to be dealt with by the conventional technique, and detect a chip along the shape to inspect it.

[0004]

According to the present invention, there is provided a defect inspection method for inspecting a defect in a contour portion by using image processing. 1) Pixels forming a binary boundary by binarizing an imaging input of an inspection object portion. Is set as a contour pixel, and 2) a reference vector is set at one end of the contour pixel toward a pixel that is a set number apart, and an inspection vector is set from each contour pixel to a pixel that is the same set number as described above. The absolute value of the angle difference representing the difference between the directions of each inspection vector and the reference vector is calculated. 3) When the absolute value of the angle difference exceeds a preset value,
The start pixel of the inspection vector is recognized as a contour shape conversion pixel, and 4) the line type between the contour shape conversion pixels recognized above is
Identification is performed based on the cumulative frequency distribution of the absolute values of the angular differences between the contour shape conversion pixels and the relationship between the preset cumulative frequency distribution shape and the line type, and 5) the identified line type and the contour shape conversion pixel Based on the coordinate values of the contour pixels in between, calculate the contour approximation line that expresses the contour shape of the inspection target by a mathematical formula, 6) overlay the contour approximation line on the contour pixel on the screen, and 7) contour approximation The pixels through which the line passes are scanned from the pixels intersecting the screen frame until the contour pixel is reached, 8) The contour pixel is scanned from the contour pixel until the contour approximate line is again reached, and 9) The contour pixel scanned is scanned. Of these, on the XY plane set on the screen, first, the pixels at the maximum and minimum positions in the X direction are extracted and X
In addition to being represented by coordinate values, a straight line passing through this point and perpendicular to the X axis is drawn, and then pixels at maximum and minimum positions in the Y direction are extracted and represented by Y coordinate values. Draw a straight line and calculate the intersections of these straight lines. 10) Extract 3 points on the inside of the inspection part from these intersections. 11) From these 3 points, close to the contour approximation line. Draw a line perpendicular to the contour approximation line, and draw a line along the contour line from the remaining points. 12) For each of the three lines, translate until they come into contact with the contour coordinates, and 13) For each 3 after translation. The feature is that the area surrounded by two lines is cut and the size is inspected.

[0005]

FIG. 1 is a block diagram showing an embodiment of the present invention.
~ It demonstrates based on FIG. FIG. 1 is a diagram showing an outline of an apparatus according to the present invention. The illumination means 2 for emphasizing the chipping of the inspection object 1, the imaging means 3 including an industrial television camera for imaging the inspection object, and the imaging means 3 are connected,
An image processing device 4 which converts an image pickup signal into binarized pixel data and stores the binarized pixel data in a memory, and displays the binarized pixel data on a screen on a monitor 7, and a memory of the image processing device 4 connected to the image processing device 4. A controller 5 having a computer for performing a chipping inspection process based on the above, and a CRT 6 connected to the controller 5 for displaying a chipping inspection result and the like are provided.

The operation of the present invention will be described according to the processing flow shown in FIG. First, the outline of the processing will be briefly described based on the processing purpose blocks 201 to 205. In the block 201, the image processing apparatus 4 stores the image pickup signal of the inspection target as binarized pixel data in the memory and displays it on the screen. After the block 202, the control device 5 mainly processes, and in the block 202, the line type forming the contour and its configuration are identified based on the binarized pixel data. In block 203, a contour approximation line in which a line type is mathematically expressed for each of the contour constituent lines and their intersections are obtained. In block 204, pixels representing the binarized contour are sequentially scanned based on the contour approximation line, and the chip and its size are calculated. Block 205 is block 204
From the size of the chip calculated in and the preset judgment value,
It is to determine whether or not it should be omitted.

The processing contents will be described below in the order of step numbers shown in FIG. In step 101, the illuminating unit 2 emphasizes the contour of the inspection object 1, the image capturing unit 3 captures an image of a predetermined inspection target portion, and an image capturing signal is input to the image processing apparatus 4. At this time, the contour shape of the inspection target portion is set within the screen. In step 102, the image processing device 4 stores the image pickup signal input in step 101 in the memory as binarized pixel data (for example, the inspected object part is 1, and the background and missing parts are 0), and An image is displayed on the monitor with the horizontal direction as the X axis and the vertical direction as the Y axis. FIG. 3 shows an example of an image of the inspection portion which is binarized and displayed on the XY coordinate plane. The following description is based on this image.

In step 103, the control device 5
The binarized data in the monitor screen is scanned one pixel at a time in the X or Y direction, and, for example, of the pixels determined to be 1, pixels in contact with pixels of 0 are extracted as contour pixels, and first, the pixels intersecting the screen frame 2 The coordinates (x ₀ , y ₀ ) and (x _n-1 , y _n-1 ) of the pixel S and the pixel E, which are the contour pixels of the location, are calculated. In step 104, the boundary pixels from the pixel S to the pixel E are scanned pixel by pixel, and the coordinate data of the central portion of a total of n contour pixels including the pixel S and the pixel E are calculated. Hereinafter, the coordinate data of the central portion of the pixel will be referred to as pixel coordinates. The coordinate data is the distance in the X and Y directions from the origin of the XY coordinates (indicated as the lower left corner in FIG. 3) set in the screen to the center position of the pixel.

In step 105, the reference vector a
(Hereinafter, it is also simply referred to as a. In FIG. 3, an arrow is shown above the symbol a.) And a plurality of check vectors bi (i = 0 to 0).
nt-1. Hereinafter, it is also simply referred to as bi. In FIG. 3, reference numeral b
Is indicated by an arrow above. ), And each bi for a
The absolute value (hereinafter simply referred to as an angle difference) θi of the angle created by is calculated. Here, a is a t-th pixel (for example, t
= 30) is a vector connecting the contour pixel coordinates (Xt, Yt). In bi, the start point of the vector is (n−t) vectors which are the contour pixel coordinates sequentially starting from the pixel S coordinate and being adjacent, and the end point of each vector is the t-th contour pixel coordinate from the start point. The a, bi and the angle difference θi can be expressed as follows based on the contour pixel coordinates. a = (Xt−X0, Yt−Y0) bi = (Xt + i−Xi, Yt + i−Yi) θi = | tan ⁻¹ ((Yt−Y0) / (Xt−X0)) − tan ⁻¹ ( (Yt
+ i−Yi) / (Xt + i−Xi)) | (where i = 0 to n−t−1), where i is 0, 1, 2, ..., Nt−1, a
(N−t) angle differences of each bi with respect to θ0, θ1, θ
2, ..., Find θn-t-1.

Next, at step 106, the angle difference θi
Based on, the conversion point of the contour shape is obtained. As shown in FIG. 4A, the abscissa plots the starting point pixel order of (nt) inspection vectors bi and the ordinate plots the angle difference θi. For example, the contour shape of the inspection portion is three types. If the angle difference θi exceeds two preset thresholds th1 and th2, the inspection vector bi
The starting point pixels J and K are set as shape conversion pixels, and their pixel coordinates (Xj, Yj) and (Xk, Yk) are shape conversion points (see FIG. 3). If the contour shape is made up of a combination of r types of shapes, (r-1) threshold values of the angle difference θi are set in the control device 5 in advance.

In step 107, the approximation coordinate data is divided from the contour pixel coordinate data in order to calculate each contour approximation line for defining each shape by a mathematical expression.
In the above description, the coordinates (Xj, Yj) and (Xk, Yk) as the shape conversion points were obtained for the three types of shapes. From this, the approximation coordinate data (Xi, Yi) for calculating the outline approximation line below is represented by the central portion of the outline pixel in the range indicated by i below. Approximate first shape ... i = 0 to j Approximate second shape ... i = j + 1 to k Approximate third shape ... i = k + 1 to n-1

In step 108, the line type of each contour approximation line is recognized, and the frequency distribution of the angle difference θi is obtained and identified from this. The control device 5 is previously provided with a line type, for example, straight line,
The relational data of the frequency distribution of the circle, the quadratic curve, etc. and the angle difference θi, and the logic for determining that it is a linear straight line when the frequency is concentrated and a circle when the frequency is flat are registered. FIG. 4B shows the frequency distribution of the change in the angle difference θi shown in FIG. In this case, i = 0 to j
, The line shape of the first shape in the range of i = (j + 1) to k is the circle of the line shape of the second shape in the range of i = (j + 1) to k.
The line shape of the third shape in the range of 1) can be judged as a straight line. An example of another contour shape is shown in FIG. 5, but FIG.
In the case of the frequency distribution of the angle difference θi such as, the line shape of the first shape in the range of i = 0 to j is a straight line, and the line shape of the second shape in the range of i = (j + 1) to k is also Straight line, i = (k + 1) to (n-1)
The line type of the third shape in the range can also be determined as a straight line. Here, the frequency around i = j and the frequency around i = k are flat, indicating that there is a circular portion. That is,
It can be determined that the straight line of the first shape and the straight line of the second shape are connected by an arc. The same applies to the second shape straight line and the third shape straight line. As described above, the line type of each contour line is automatically identified. The relationship between the contour shape and the line type may be registered in advance and this information may be used.

In step 109, each contour shape from the first shape to the third shape is calculated by using the identified line type and the approximation coordinate data, and the respective contoured approximation lines L1 and contours are calculated. An approximate line L2 and a contour approximate line L3 are calculated (hereinafter, also simply referred to as L1, L2, L3). In the contour shape shown in FIG. 4, as described above, L
1 and L3 are linear least-squares approximations, and L
2 is calculated using the least squares approximation of the circle equation. In step 110, the contour approximation line L obtained in step 109
An intersection p1 between the screen frame and L1, an intersection p2 between L1 and L2, an intersection p3 between L2 and L3, and an intersection p4 between L3 and the screen frame are obtained based on the mathematical expressions representing 1, L2, and L3. FIG. 6 shows the relationship between the binarized image and L1, L2, L3 based on the binarized image and the above-described intersection points p1, p2, p3, p4.

The method of detecting and inspecting a chip will be described below with reference to steps 111 to 121. In step 111, the pixels including the contour approximation line are scanned along the contour approximation line, and the pixel U immediately before changing from 1 to 0 is extracted. The pixel U is a defect detection start pixel for detecting a defect by scanning the binarized contour pixel. As the contour approximation line, first, with respect to L1, the area between intersection points p1 and p2 is processed. If there are a plurality of missing detections during this period, there will be a plurality of missing detection start pixels, and the later-described missing detection processing is performed for each missing detection start pixel. In step 112, using the pixel U as a starting point, scanning is performed on the contour pixel with respect to the contour approximation line L1 in the inward direction of the inspection object until a pixel at which the contour approximation line L1 passes again is reached, and the scanned contour pixel coordinates are determined. Ask. FIG. 7 is an enlarged view of the range from the intersection p1 to the intersection p2 in FIG. 6, which will be described below with reference to FIG. In step 113, the maximum value X _max and the minimum value X in the X direction obtained from the scanned contour pixel coordinate data.
_min , and the maximum value Y _max and the minimum value Y _min in the Y direction are obtained. In step 114, a straight line perpendicular to the X axis passing through the maximum value X _max and the minimum value X _min in the X direction and a straight line perpendicular to the Y axis passing through the maximum value Y _max and the minimum value Y _min in the Y direction are drawn.
The points A, B, and C located inside the object to be inspected are obtained from the four intersections of the four straight lines. That is, the rectangle parallel to the XY axis having the points A, B, and C as vertices contains a chip,
It can represent the size of the chip. In FIG. 7, the coordinates of points A, B, and C are as follows. A (X _min , Y _max ) B (X _max , Y _min ) C (X _max , Y _max )

In step 115, the defect detection auxiliary lines v1, v2 and h are obtained. v1 is a straight line passing through the point A and perpendicular to the contour approximation line L1, v2 is a straight line passing through the point B and perpendicular to L1, and h is a line passing through the point C and along L1. Step 1
At 16, the intersection of the lines L1, v1, v2, and h is obtained.
The figure connecting the intersections is a rectangle or a quadrangle with one side of the contour approximation line, and for converting the evaluation of the size of the chip from a reference parallel to the XY axes to one based on the contour approximation line. It is provided in. In step 117,
First, v1 is scanned from the intersection with L1 to the intersection with h to check whether or not there is a contour pixel. When the contour pixel is not detected up to the intersection with h, v1 is shifted by one pixel to the side with a defect to form a new straight line v1, and scanning is performed again from the intersection with L1 to the intersection with h. Check for the presence of contour pixels. This process is repeated until the contour pixel is detected. In step 118, v1 when the contour pixel is detected is calculated as the missing detection line v1a.

Step 119 is an iterative loop for performing the processing of steps 117 and 118 on the remaining defect detection auxiliary lines v2 and h. Therefore, when step 119 is exited, as shown in FIG. 7, with respect to the defect detection auxiliary lines v2 and h, the defect detection lines v2a and ha are calculated in the same manner as in the case of v1. Step 12
At 0, the calculated vertical distance W between v1a and v2a and the vertical distance D between L1 and ha are calculated. Let W and D be the width and depth of the chip existing in the inspection portion. Therefore, no matter how the contour of the inspection object is tilted on the screen, the width and depth of the chip can be calculated without being affected by the tilt. This value is based on the actual inspection based on the contour.

Step 121 includes steps 111 to 12
This is a loop for performing the operation of 0 for L2 and L3. Therefore, when step 121 is skipped, L2 and L3
The width W and the depth D of the chip in the range of are required. L2
In contrast, the same chipping detection processing as described above is performed based on the contour approximation line L2 that substantially follows the arc-shaped second shape of the inspection object. As described above, if there are a plurality of missing detection start pixels, the above processing is performed for each pixel. In step 122, the width W and the depth D are compared with a predetermined determination value for all the defects obtained up to step 121. If the width W and the depth D are larger than the determination values, the inspection object has a defect. The determination process for determining is performed. The lack determination result can be output to the outside by appropriately displaying it on the CRT 6.

[0018]

As described above, the present invention has the following effects. Even for an inspected object with a complicated shape whose contour shape cannot be expressed by a single mathematical expression, if only the information about the number of combinations of shapes is set, the contour approximation line that approximates the actual contour shape can be set automatically. Since it is not necessary to register information about all the line types forming the contour shape in detail and their positions and the like, it is possible to easily deal with the change of the contour shape of the inspection object, so that the application range of the chipping inspection is expanded. In addition, since the width and length of the chip are measured with reference to the contour approximation line, not with the reference parallel to the XY axes of the imaging screen, the contour of the object to be inspected on the screen may be inclined with respect to the XY axes. However, the size of the chip is calculated equally, and the chip inspection with high reliability can be performed. Further, since the binarized contour pixels imaged on the basis of the contour approximation line are scanned to detect and inspect for defects, only the contour shape portion can be subjected to the defect inspection, so that efficient defect inspection can be performed.

[Brief description of drawings]

FIG. 1 is a diagram showing a device configuration for explaining an embodiment.

FIG. 2 is a diagram showing a flow for explaining an embodiment.

FIG. 3 is a conceptual diagram showing the relationship between contour pixels, basic vectors, and inspection vectors.

FIG. 4 is an example showing an absolute value of a difference between an inspection vector position and a direction with respect to a basic vector.

FIG. 5 is another example showing the absolute value of the difference between the inspection vector position and the direction with respect to the basic vector.

FIG. 6 is a diagram illustrating a method of detecting a chip.

FIG. 7 is a diagram illustrating a method of calculating the size of a chip.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Inspected object 3 ... Imaging means 4 ... Image processing device 5 ... Control device a ... Basic vector b ... Inspection vector θ ... Absolute value of angle formed by a and b L1, L2, L3 ... Contour approximation line U ... Missing Detection start pixel v1a, v2a, ha ... A line that surrounds the chip on the basis of the contour approximation line W ... Chip width D ... Chip depth

Claims (3)

[Claims] 1. A defect inspection method for inspecting a defect in a contour portion by using image processing, comprising: 1) binarizing an imaging input of a portion to be inspected and defining pixels forming a binary boundary as contour pixels; 2) contour At one end of the pixel, a reference vector is set toward a pixel that is a set number apart, and an inspection vector is set from each contour pixel to the same set number of pixels apart from the above, and each inspection vector is set to the reference vector. And 3) when the absolute value of the angle exceeds a preset value, the start pixel of the inspection vector is recognized as a contour shape conversion pixel, and 4) the contour shape recognized above. The line type between the converted pixels is
Identification is performed based on the cumulative frequency distribution of the absolute values of the angles between the contour shape conversion pixels and the relationship between the preset cumulative frequency distribution shape and the line type, and 5) the identified line type and the contour shape conversion pixel Based on the coordinate values of the contour pixels in, the contour approximation line expressing each contour shape of the inspection target by a mathematical formula is calculated, and 6) the chipping inspection process is performed along the contour approximation line. Chip inspection method. 2. A defect inspection method for inspecting a defect in a contour portion by using image processing, comprising: 1) binarizing an imaging input of an inspection object portion, and pixels forming a binary boundary are contour pixels; and 2) contour The contour approximation line is calculated based on the line type information that forms the contour shape with the pixel, and 3) the contour approximation line is superimposed on the contour pixel on the screen, and 4) the pixels through which the contour approximation line passes are displayed on the screen frame. From the pixel intersecting with the contour pixel, scanning is performed until the contour pixel is reached. 5) The contour pixel is scanned until the contour approximation line is again reached from the contour pixel. 6) Among the scanned contour pixels, first, in the XY plane set on the screen, The pixels at the maximum and minimum positions in the X direction are extracted and X
In addition to being represented by coordinate values, a straight line passing through this point and perpendicular to the X axis is drawn, and then pixels at maximum and minimum positions in the Y direction are extracted and represented by Y coordinate values. Draw a straight line and calculate the intersections of these straight lines. 7) Extract 3 points on the inside of the inspection part from these intersections. 8) From these 3 points, close to the contour approximation line. Draw a line perpendicular to the contour approximation line, and draw a line along the contour line from the remaining points. 9) For each of the three lines, move in parallel until they touch the outline coordinates, and 10) For each 3 after the parallel movement, A chipping inspection method characterized in that the area surrounded by two lines is chipped and the size is inspected. 3. A defect inspection method for inspecting a defect in a contour portion by using image processing, comprising: 1) binarizing an imaging input of a portion to be inspected, a pixel forming a binary boundary being a contour pixel, and 2) a contour. At one end of the pixel, a reference vector is set toward a pixel that is a set number apart, and an inspection vector is set from each contour pixel to the same set number of pixels apart from the above, and each inspection vector is set to the reference vector. And 3) when the absolute value of the angle exceeds a preset value, the start pixel of the inspection vector is recognized as a contour shape conversion pixel, and 4) the contour shape recognized above. The line type between the converted pixels is
Identification is performed based on the cumulative frequency distribution of the absolute values of the angles between the contour shape conversion pixels and the relationship between the preset cumulative frequency distribution shape and the line type, and 5) the identified line type and the contour shape conversion pixel Based on the coordinate value of the contour pixel in, the contour approximation line that represents the contour shape of the inspection target by a mathematical formula is calculated, and 6) the contour approximation line is superimposed on the contour pixel on the screen, and 7) the contour approximation line. The pixels passing through are scanned from the pixels intersecting the screen frame until the contour pixel is reached, 8) The contour pixel is scanned until the contour approximate line is again reached from the contour pixel, and 9) Of the scanned contour pixels , First, in the XY plane set on the screen, the pixels at the maximum and minimum positions in the X direction are extracted and X
In addition to being represented by coordinate values, a straight line passing through this point and perpendicular to the X axis is drawn, and then pixels at maximum and minimum positions in the Y direction are extracted and represented by Y coordinate values. Draw a straight line and calculate the intersections of these straight lines. 10) Extract 3 points on the inside of the inspection part from these intersections. 11) From these 3 points, close to the contour approximation line. Draw a line perpendicular to the contour approximation line, and draw a line along the contour line from the remaining points. 12) For each of the three lines, translate until they come into contact with the contour coordinates, and 13) For each 3 after translation. A chipping inspection method characterized in that the area surrounded by two lines is chipped and the size is inspected.