JPH02242382A - Defect checking method - Google Patents

Abstract

PURPOSE:To easily detect a defective part regardless of a low contrast by detecting an edge flag point which can be regarded as a point on the edge of the defective part and detecting the change of density of an original picture around the edge flag point. CONSTITUTION:The image of a check object is picked up by a picture input device 1 like a television camera, and the density of each picture element is converted to a digital signal by an A/D converting part 2 and is subjected to preprocessing in a preprocessing part 3 to obtain a differential absolute value picture, a differential direction value picture, and an edge picture besides the original picture obtained from the converting part 2. Points indicating that a larger differential absolute value corresponds to a larger density change are noticed to perform the line narrowing processing. That is the differential absolute value of each picture element is compared with those of nearby picture elements to extract an edge having the width corresponding to one picture element.

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to a defect inspection method for inspecting external defects such as chips, cracks, dirt, and foreign objects on the surface of an object to be inspected by image processing.

[Conventional technology]

Conventionally, a spatial region containing an object to be inspected is imaged using an image input means such as a television camera, the density of each pixel of the original image is binarized using an appropriate threshold, and defective parts are identified based on this binary image. For example, there are known methods for detecting
2-88946). That is, if there is a region in the binary image where the values are inverted inside the region that can be considered as the inspection object, it is regarded as a defective region.

[Problem M that the invention attempts to solve]

In the above conventional method, a defective part is determined based on a binary image. (is above the threshold value), there is a problem that the defective part cannot be identified. The present invention aims to solve the above problem, and detects edge flag points that can be regarded as points on the edge of a defective part, and detects density changes in the vicinity of the edge flag points or the vicinity of each pixel on the contour line of the defective part. The present invention aims to provide a defect inspection method that improves the accuracy of detecting defective parts by detecting changes in the concentration of .

[Means to solve the problem]

In the present invention, in order to achieve the above object, at least one search line is set inside the outline of the object to be inspected for the edge image, and when an edge flag point that can be regarded as an edge of a defective part is detected on the search line, Defects are identified based on density changes in the original image near edge flag points. In addition, for edge images, at least one inspection area is set inside the outline of the object to be inspected, and when an edge flag point that can be regarded as an edge of a defective part is detected by scanning and searching the entire area of the inspection area, the edge flag is The outline of the defective part may be traced using a point as a starting point, and the defective part may be identified based on density changes in the original image near each pixel on the outline.

[Effect]

According to the first method, an edge flag point that can be considered as a point on the edge of a defective part is detected, and a density change in the original image in the vicinity of the edge flag point is detected. Therefore, even if the defective part is defective, it can be detected, and the accuracy of detecting the defective part is improved. Also, in the second method, since density changes in the vicinity of the contour of the defective portion are detected, even defective portions with relatively low contrast can be detected.

[Embodiment 1] In the present invention, as shown in FIG. 1, an image of an object to be inspected is captured by an image input device 1 such as a television camera, and the density of each pixel is converted into a digital signal by an analog-digital converter 2. Thereafter, the preprocessing section 3 performs the following preprocessing to obtain, in addition to the original image obtained from the analog-to-digital conversion section 2, a differential absolute value image, a differential direction value image, and an edge image. That is, the original image P0 obtained by imaging a spatial region including the inspection object is a grayscale image, and as shown in FIG. 2(a), it is an image containing inspection object 0, defect X1, foreign matter It becomes. The process of extracting edges such as the outline of the inspection object 0 from this grayscale image is based on the idea that edges correspond to areas with large density changes. Therefore, the differentiation process that extracts edges by differentiating the density is as shown in Figure 3.
The original image P0 is divided into locally parallel windows W of 3 x 3 pixels. In other words, the pixel E of interest and the 8 pixels (8 neighbors) around that pixel E are locally paralleled. form a window W, and a pixel A within the locally parallel window W
Find the vertical density change V and the horizontal density change ΔH of the density of ~■ by the following formula, ΔV=(A+B〇)−(G+H+I)ΔH=(A+D
+G)-(C+F+I) Furthermore, the differential absolute value 1e11 and the differential direction value le t are determined by the following equation. el-=(muV2+ΔH2)I/2 1 e g = ta□-・αyay ΔH However, A to A indicate the density of the corresponding pixel. As is clear from the above equation, the absolute differential value 1e represents the maximum rate of change in density in the vicinity of the pixel of interest in the original image, and the differential direction value l e t represents the direction of the maximum change in density in the vicinity of the pixel of interest. By performing calculations of 0 or more representing a direction perpendicular to , and the direction of the change. Here, an image in which the density of each pixel is expressed by a differential absolute value let is called a differential absolute value image, and an image expressed by a differential direction value let is called a differential direction value image. Next, a thinning process is performed. The thinning process is performed by focusing on the point that the larger the absolute differential value, the larger the density change. That is, by comparing the differential absolute value of each pixel with the differential absolute values of surrounding pixels and connecting those that are larger than the surrounding pixels, an edge with a width of one pixel is extracted. As shown in Figure 4, the position of each pixel is expressed by X-Y coordinates, and the differential absolute value is expressed by Z.
Considering the differential absolute value image P1 taken along the axis, the thinning process corresponds to finding the edge line on this curved surface. Through the processing up to this point, regardless of the magnitude of the absolute differential value,
All edges are extracted. The ridge line obtained at this stage also includes unnecessary small peaks due to noise, etc., so as shown in Figure 4,
A threshold value SL is appropriately set, and only values equal to or higher than the threshold value SL are employed to remove noise components. The image obtained by this process tends to have discontinuous lines when the contrast of the original image is insufficient or there is a lot of noise. Therefore, edge extension processing is performed. Starting from the end point of the line, the pixel of interest and its surrounding pixels are compared, and an evaluation function f(
Extend the line in the direction where ea) is the largest, and continue this until it collides with the end point of another line. f(e4)=3 Secondly, eo is the differential data of the central pixel (corresponding to E of the local parallel window W), and e is the differential data of the adjacent pixel (corresponding to the 8 neighbors of the local parallel window W) and j=1.2.・・・・・・It is 8. Through the above processing, as shown in FIG. 2(b), an edge image P is obtained that traces a portion of the original image where the density change is large. In the edge image P4, the original image, differential absolute value image, differential direction code image, edge Four types of images are obtained, and each image is stored in an original image memory 4, a differential absolute value image memory 5, a differential direction value image memory 6, and an edge image memory 7, respectively. In the following explanation, it is assumed that the position of a pixel in each image is expressed by X-Y coordinates. Let the density of each pixel in each image be f+(x, y>,
Let f2 (x, y), f3 (x, y), and f4 (x, y). The original image is a grayscale image, and the density is usually represented by 8 bits, so the density at each pixel is a(=f(x,y))
is 0≦a≦255. Further, the density of the differential absolute value image (i.e., the differential absolute value) b (= fz (x, y)) is expressed by, for example, 6 bits, and 0≦b≦63, and the density of the differential direction value image (i.e., Differential direction value) c (=
f 3 (X, y)) is expressed, for example, in 16 directions, and satisfies 0≦C≦15. For an edge image, since only the presence or absence of a line is present, pixels forming a line are represented as "1n", and other pixels are represented as "0". That is, fn(x,
The value range of y) is (0, 1). In the following explanation, the term density represents the density of white, and the larger the density value, the brighter it is. Next, as shown in FIG. 7, the search processing section 6 sets a search line S inside the contour line of the edge image P4. The search line S can be arbitrarily set as long as it has a closed loop shape, as shown in FIGS. 8(a) to 8(d). That is, a search line S whose starting point and ending point coincide is set. After setting the search line S, start the search and find a point on the search line S where f4 (x, y) = l; 1, then set this point as the edge flag point 0
Next, as shown in FIG. 9, in the stick mask setting section 9, starting from the starting point, centering on the pixel of the first edge flag point F found, a plurality of pixels (number 1) before and after the edge flag point F on the search line S are 8) are extracted and the stick mask M is set. The stick mask M is an area with a width of 1 and a length of 2n+1, where:
n is a natural number and is set appropriately. Now let the coordinates of the edge flag point F be (xr, yy), and the coordinates of each pixel in the stick mask M be RjXr+n+3' r+n)+
""...+R+(Xr+++3/Fl), F (X
r, 3'F), SI(XF-1,'/r-+
), '・”・', Sh (X F-n + yF-n
>, the density calculation unit 10 calculates the following formula to find the sum of differences in density of pixels on the original image P0 located at symmetrical positions across the edge flag point F within the stick mask M. D=ΣIf 1(XF+l,yF-1) f+(X
r-+, 3'r-+) where i=1.2.・・・・・・
., n, or i=m. m+1. ..., n (m≠1). Judgment part 1
1, the sum obtained as above is compared with a predetermined threshold SL1, and if D>SLI, the search line S
It is determined that there is a defective part such as a chip or a foreign object on the top. In other words, if the edge flag point F is a point on the contour line of the defective part, the stick mask M will straddle the defective part and a part that is not a defective part, so the edge flag point F
By finding the density difference between pixels located at symmetrical positions with F as the center, the contrast in the vicinity of the edge flag point F can be detected. Furthermore, since density differences are detected for a plurality of pixels, contrast is emphasized and detection accuracy can be improved. When D≦SL1, the search on the search line S is continued to detect the presence or absence of an edge flag point. In this way, when the end point of the search line S is reached and no edge flag point is detected, it is determined that no defective portion exists.The flow of one or more processes is shown in FIG.

[Embodiment 2] In the above embodiment, the search line S was formed into a closed loop shape, but in this embodiment, as shown in FIG. It is located at Here, the search line S can be arbitrarily set in shape, such as a straight line, a curved line, or a broken line, as long as it is inside the outline of the object to be inspected, and its direction can also be set arbitrarily. The flow of the process is shown in FIG. 10. After setting the search line S, the edge flag point F is detected in the same manner as in the first embodiment, and the edge flag point F is detected in the same manner as in the first embodiment. A stick mask M is set using the following method, and the sum of density differences between pixels existing at symmetrical positions on the stick mask M is determined. If this sum is D>SL for a predetermined threshold SL1,
If I is satisfied, this edge flag point F is designated as a defect candidate point. However, in this embodiment, the edge flag point F is D>S
Rather than immediately determining that it is a defective part even if LI is satisfied, after calculating the total number K of such defective candidate points on the search line S, the total number K of defective candidate points is set to a predetermined threshold value SL2. By comparison, when K>SL2, it is determined that a defective part exists.6 In other words, the conditions for determining whether a defective part exists are made stricter than in Example 1, and the detection accuracy is further improved. This is possible.

Embodiment 3 In this embodiment, as shown in FIG. 13, a plurality of search lines 81 to So having the same shape are set. In this case, the 14th
As shown in the figure, if the defective part X1 exists, the defective part X1 is found consecutively on multiple search lines. When the search line becomes a defect candidate line and the total number of defect candidate lines is compared with a predetermined threshold value SL3 and L>SL3, it is determined that the defective part X1 exists. Here, the determination of whether the defective part X1 exists on one search line is as follows:
Embodiment 1 The process flow of this embodiment is as shown in FIG. 12, and except for the part indicated by the broken line in the figure, the defective part This means that it was detected using the method of Example 1, and if the part indicated by the broken line is included, it means that it was detected using the method of Example 2.

[Embodiment 4] In each of the above embodiments, the search line was set inside the outline of the object to be inspected, but in this embodiment, as shown in FIG. A reference density setting area DM and an inspection area DM are set. The flow of processing is as shown in FIG. 15, and will be explained below. In the reference density setting area DM, a reference density SD, which is the average value of the density of all pixels in the reference density setting area DM, is determined from the original image. Next, regarding the inspection area D M 2, the 17th
As shown by the arrow in the figure, rask scanning is performed and f 4 (X
, y ) = l. Let the coordinates of the edge flag point F0 first found by rask scanning be (x o , yo), and this edge flag point F
. As a starting point, as shown in FIG. 18, each edge flag point F is sequentially connected to the edge flag point Fo (Xo, :Yo). In the following explanation, the coordinates of each edge flag point F are assumed to be (XK, 3/K), where n=1.2. ...is... Assume that processing is now being performed on edge flag point F1. The edge flag point F t(x l
, y +), a differential direction value S is obtained from the differential direction value image, and a stick mask M is set in a direction (normal direction) orthogonal to the direction indicated by the differential direction value S. As shown in FIG. 19, the stick mask M is set to have a symmetrical shape with the edge flag point F as the center.
Pixels in the stick mask M do not need to be in contact; here, a pixel with n pixel layers and a pixel with m pixel layers from the edge flag point Ft (xl, y□) are assumed to be pixels in the stick mask M. There is. The coordinates of each point of the stick mask M are (X I ell 13' t +j, (X l
*n + 3' 1on), (X +-n+y+-n
), (X + - m + 3' l - J, find the density value of each point on the original image, f + (X t, m, 3' I + j = d + f + ('
)C+*n,'j 11111)=d 2f +(X
+-m+3't-j=dcof I(X r-+a,
3' 1-)=d4. Next, the inspection area DM
, the defective part X, existing in In the example,
In order to determine whether it is a black defect (hereinafter referred to as a black defect), the density values d1 to d4 in the stick mask M are compared with the reference density SD. Here, dl. d2. If the condition di, d4>5D (all of d4 to dl is greater than SD) is satisfied, in the subsequent processing,
A white threshold WSL is used, and if the conditions are not satisfied, a black threshold BSL is used. The white threshold value WSL and the black threshold value BSL may be appropriately set depending on the contrast of the defective portion x1. That is, when the conditions for using the white threshold value WSL are satisfied, for the density difference between the symmetric points in the stick mask M, l d + d 41 > W S L,
! It is checked whether each condition of d 2 d s I > W SL is satisfied, and if any one of them is satisfied, the density difference that satisfies the condition is set as the white addition value. Similarly, if the conditions for using the black threshold BSL are satisfied, ld+d<I>BSL, 1d2
It is checked whether each condition -djl>BSL is satisfied, and if any of the conditions is satisfied, the density difference that satisfies the condition is set as the black addition value. After obtaining the white addition value and the black addition value as described above, the presence or absence of an edge flag point adjacent clockwise from edge flag point F is checked. If there is an adjacent edge flag point, perform the above processing, and then check whether there is an adjacent edge flag point and perform tracking. While tracking, calculate the sum of the white addition value and the black addition value. . This tracking ends when one of the following stopping conditions is met: ■When the number of pixels to be tracked exceeds the preset maximum tracking number Tm; ■When adjacent edge flag points leave the inspection area DM2; ■When there are no adjacent edge flag points. After tracking is completed, The total sum TW of the white added value and the total sum TB of the black added value are compared with a preset threshold value W for the white defect portion and a threshold value B for the black defect portion, respectively, and TW>W.
, it is determined that a white defect exists in the inspection area DM2, and when T B > B + is satisfied, it is determined that a black defect exists in the inspection area DM2. . Moreover, when any of the conditions is not satisfied, it is determined that there is no defective part within the inspection area DM2. Each threshold value WSL, BSL, W, ,B,, maximum tracking number T
m may be appropriately set depending on the size and contrast of the defective portion.

[Embodiment 5] In this embodiment, as shown in FIG. 21, only the inspection area DM is set within the outline of the object to be inspected, and this inspection area DM
A rask scan is performed within the area to detect the edge flag point, which is the starting point for tracking. The processing procedure is as shown in FIG. 20, in which the stick mask M is set as in Example 4,
The densities d1 to d4 of each pixel in the stick mask M are determined. The density difference of the symmetrical point in the Sarani stick mask M is determined and compared with a predetermined threshold value SL. That is, ld
Edge flag points are tracked while checking whether at least one of the following conditions is satisfied: +d4>SL+Id2d:l1>SL+. During tracking, the number of edge flag points that satisfy the conditions is counted. Further, the differential absolute value of the tracked edge flag point is determined based on the differential absolute value image, and the differential absolute values are sequentially added. The stopping conditions for tracking edge flag points are the same as in the fourth embodiment. Once the tracking is finished,
Since the sum of the edge flag points that satisfy the above conditions and the sum of the differential absolute values for the edge flag points are calculated, a preset threshold value SL2. S
Compared with L, K>SL. L>SL. When at least one of the conditions is satisfied, the inspection area DM,
It is determined that there is a defective part within. When no defective part is found, the above-mentioned processing is performed on the part of the inspection area DM that has not been scanned, and when no edge flag point is found, or the threshold value S L, , S L
, it is determined that no defective portion exists within the inspection area DM.

[Embodiment 6] In this embodiment, as in Embodiment S4, the reference density setting area DM,
and the inspection area D M 2, and set the reference concentration setting area D.
The average value of the densities within M+ is set as the reference density SD. The following processing is as shown in FIG. 22, and the defective area with a density value brighter than this standard density SD (hereinafter referred to as a white defective area in this embodiment) and the defective area with a darker density value (
In this embodiment, in order to detect a black defect (hereinafter referred to as a black defect), a white threshold WSL for detecting a white defect and a black threshold BSL for detecting a black defect are set. That is,
As shown in FIG. 23, bright and dark offset values LT and DK are set for the reference density SD, respectively, and a white threshold WSL and a black threshold BSL are set as shown in the following equations. WSL-3D+LT BSL=SD-DK In this way, white threshold WSL and black threshold BSL
After setting, the inspection area DM,
The edge flag point, which is the starting point of tracking, is detected. When the edge flag point F is detected, the 24th
As shown in the figure, density measurement points P R and P t are set at symmetrical positions across the edge flag point in a direction orthogonal to the differential direction value at that position. Here,
Concentration measurement point P1. l. Pt is set at a position separated by n pixels from the edge flag point Fi. As is clear from the definition, the differential direction value has the property that on the original image, the left side of the direction indicated by the differential direction value is dark and the right side is bright. The density U in the original image is compared with the white threshold value WSL, and the density measurement point P on the left is
The density V in the original image of L is compared with the black threshold BSL. Here, if the condition u>WSL is satisfied, that edge flag point is set as a white defect candidate point. Also, v<BSL
If it satisfies the following, the edge flag point is designated as a black defect candidate point. This process is repeated while tracking a series of connected edge flag points, and the total number of edge flag points regarded as white defect candidate points and edge flag points regarded as black defect candidate points is determined. The conditions for stopping tracking of edge flag points are the same as in the fourth embodiment. The total number of white defect candidate points at the time of finishing the tracking is set as SW, and the total number of black defect candidate points is set as SB, and these are compared with preset threshold values SLv and SL4. That is, S
W>SL 3 SB>SL. If at least one of these conditions is satisfied, it is determined that there is a defective part in the inspection area 0M2, and if neither of these conditions is satisfied, a rask scan is performed on the part of the inspection area DMf that has not yet been rask scanned, and the same process is performed. Inspect the defective part using the following procedure. When the rask scan is completed for the entire area in the inspection area 0M2, if there is no edge flag point or if both conditions regarding the thresholds S L, , S L are not satisfied, there is a defective part in the inspection area 0M2. It is assumed that does not exist.

【Effect of the invention】

As described above, the present invention sets at least one search line inside the contour line of the object to be inspected for an edge image, and when an edge flag point that can be considered as an edge of a defective part is detected on the search line, This system identifies defective areas based on changes in density in the original image at the edge of the defective area. Therefore, even a defective portion having a relatively small contrast with the surroundings can be detected, and there is an advantage that the accuracy of detecting the defective portion is improved. In addition, for edge images, at least one inspection area is set inside the outline of the object to be inspected, and when an edge flag point that can be regarded as an edge of a defective part is detected by scanning and searching the entire area of the inspection area, the edge flag is If the outline of the defective part is traced using the point as a starting point, and the defective part is identified based on the density change in the original image near each pixel on the outline, the outline of the defective part can be traced. Because it detects density changes, even defective areas with relatively low contrast can be detected.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a processing circuit corresponding to the method of Embodiment 1 of the present invention, FIG. is an explanatory diagram showing locally parallel windows in the same as above, the fourth
The figure is an explanatory diagram showing the differential absolute value image in the same as above, Figure 5 is an explanatory diagram of the process of removing noise at the stage of obtaining an edge image from the differential absolute value image in the same as above, and Figure 6 shows the processing plain in the same as above. Fig. 7 is an explanatory diagram of the operation showing the setting state of the search line for the same edge image as above;
Figures (a) to (d) are explanatory diagrams showing the form of the search line in the same as above, Fig. 9 is an explanatory diagram of operation showing the concept of edge flag points in the same as above, and Fig. 1O is a processing in the second embodiment of the present invention. Operation explanatory diagram showing the procedure, Figures 11(,) to (d)
) are explanatory diagrams showing the form of the search line in the same as above, FIG. 12 is an explanatory diagram of the operation showing the processing procedure in the third embodiment of the present invention, FIG. The figure is the same operation explanatory diagram, No. 15.
The figure is an operation explanatory diagram showing the processing procedure in Embodiment 4 of the present invention, FIG. 16 is an explanatory diagram of the same as above, FIG. 17 is an explanatory diagram of rask scanning operation in the same as above, and FIG. 18 is a stick mask setting in same as above. Operation explanatory diagram showing the state, No. 19
20 is an explanatory diagram of the operation showing an example of the stick mask in the same as above, FIG. 20 is an explanatory diagram of the operation showing the processing procedure in the fifth embodiment of the present invention, FIG. 23 is an operation explanatory diagram showing the processing procedure in Embodiment 6 of the present invention, FIG. 23 is an operation explanatory diagram showing the relationship between the white threshold value, black threshold value, and offset value in the same as above, and FIG. 24 is the density diagram in the same as above. It is an operation explanatory diagram showing measurement points. 10-12...Contour line, x2...Foreign object. O...Inspection object,

Claims (8)

[Claims] (1) After capturing an image of a spatial region including the object to be inspected by the image input means, converting it into an edge image including the outline of the object to be inspected based on the density of each pixel of the original image obtained by the image input means; In a defect inspection method that inspects defects such as chips, cracks, dirt, etc. that exist inside the outline of an object to be inspected based on an original image and an edge image, At least one search line is set in the search line, and when an edge flag point that can be considered as the edge of a defective part is detected on the search line, the defective part is identified based on density changes in the vicinity of the edge flag point in the original image. Characteristic defect inspection method. (2) The above edge image is obtained by thinning processing based on a differential absolute value image in which each pixel is associated with a differential absolute value representing the maximum rate of change in density value in the neighborhood area of each pixel of the original image. When an edge flag point is detected on a search line set within the outline of the object to be inspected based on the image, a stick mask consisting of multiple pixels centered on the edge flag point is set on the search line, and the original image is 2. The defect inspection method according to claim 1, wherein the total sum of density differences of pixels at symmetrical positions in the stick mask is calculated based on the above, and when the sum exceeds a predetermined value, it is determined that a defective portion exists. (3) The above edge image is obtained by thinning processing based on a differential absolute value image in which each pixel is associated with a differential absolute value representing the maximum rate of change in density value in the neighborhood area of each pixel of the original image. When an edge flag point is detected on a search line set within the outline of the object to be inspected based on the image, a stick mask consisting of multiple pixels centered on the edge flag point is set on the search line, and the original image is Based on this, the sum of the density differences of pixels at symmetrical positions in the stick mask is calculated, and if the sum exceeds a predetermined value, that edge flag point is designated as a defect candidate point, and the total number of defect candidate points on the search line is set to a predetermined value. 2. The defect inspection method according to claim 1, wherein it is determined that a defective portion exists if the value exceeds . (4) For the edge image, a plurality of parallel search lines are set within the outline of the object to be inspected, and each search line is detected by the defect inspection method of claim 2 or claim 3 if a defective portion exists. Once determined, the search line is set as a defect candidate line, and multiple adjacent search lines are consecutively set as defect candidate lines, and if the total number of defect candidate lines exceeds a predetermined value, it is determined that a defective part exists. A defect inspection method characterized by: (5) After capturing an image of a spatial region including the object to be inspected by the image input means, converting it into an edge image including the outline of the object to be inspected based on the density of each pixel of the original image obtained by the image input means; In a defect inspection method for inspecting defects such as chips, cracks, dirt, etc. that exist inside the outline of an object to be inspected based on an original image and an edge image, at least When one inspection area is set and an edge flag point that can be considered as the edge of the defective part is detected by scanning and searching the entire area of the inspection area, the outline of the defective part is traced using the edge flag point as the starting point. Additionally, a defect inspection method is characterized in that a defective portion is identified based on a density change in the vicinity of each pixel on the contour line. (6) The above edge image is obtained by thinning processing based on a differential absolute value image in which each pixel is associated with a differential absolute value representing the maximum rate of change in density value in the neighborhood area of each pixel of the original image, and inspected. A reference density setting area and an inspection area are set inside the outline of the object, and the average value of the density of pixels in the reference density setting area for the original image is determined as the reference density. When a flag point is detected, the above maximum change in density value is determined based on a differential direction value image in which each pixel is associated with a differential direction value that reflects the direction of maximum change in density value in the neighborhood area of each pixel in the original image. Set a stick mask consisting of multiple pixels centered on the edge flag point in the direction of
When the density of all pixels in the stick mask for the original image exceeds the above reference density and the density difference between pixels at symmetrical positions in the stick mask exceeds a predetermined white threshold value, that density difference is used as the white addition value. , if the density of at least one pixel in the stick mask for the original image is below the reference density, and the density difference between pixels at symmetrical positions in the stick mask exceeds a predetermined black threshold, the density difference is After tracking all adjacent edge flag points and detecting a white addition value and a black addition value for each edge flag point, it is determined that at least one of the white addition value and the black addition value corresponds to each edge flag point. Claim 5 characterized in that it is determined that a defective portion exists when a predetermined value is exceeded.
Defect inspection method described. (7) The above edge image is obtained by thinning processing based on a differential absolute value image in which each pixel is associated with a differential absolute value representing the maximum rate of change in density value in the neighborhood area of each pixel of the original image, and inspected. When an inspection area is set inside the contour line of the object and an edge flag point is detected during scanning within the inspection area, the differential direction that reflects the direction of the maximum change in density value in the neighborhood area of each pixel of the original image is set. Based on the differential direction value image in which the value corresponds to each pixel, a stick mask consisting of multiple pixels centered on the edge flag point is set in the direction of the maximum change in the density value, and the symmetry within the stick mask is set for the original image. When the density difference between pixels at a position exceeds a predetermined value, that edge flag point is designated as a defect candidate point, and after tracking all adjacent edge flag points, the total number of defect candidate points and the differential absolute value of all defect candidate points are determined. 6. The defect inspection method according to claim 5, wherein it is determined that a defective portion exists when at least one of the total sum and the total sum exceeds a predetermined value set correspondingly to each of them. (8) The above edge image is obtained by thinning processing based on a differential absolute value image in which each pixel is associated with a differential absolute value representing the maximum rate of change in density value in the neighborhood area of each pixel of the original image, and inspected. A reference density setting area and an inspection area are set inside the outline of the object, and a white threshold value that is greater than the average density of pixels in the reference density setting area determined from the original image and a black threshold value that is smaller than the above average value are set. When an edge flag point is detected during scanning in the inspection area, a differential direction value is assigned to each pixel that reflects the direction of the maximum change in density value in the neighborhood of each pixel in the original image. Based on the differentiated direction value image, a pair of density measurement points separated from the edge flag point are set at symmetrical positions centering on the edge flag point in the direction of the maximum change in density value, and the density value is determined for the original image. If the density value of the larger density measurement point exceeds the white threshold, it is considered a white defect candidate point, and if the density value of the smaller density measurement point is smaller than the black threshold, it is considered a black defect candidate point, and the adjacent After tracking all edge flag points and detecting white defect candidate points and black defect candidate points for each edge flag point, at least one of the total number of white defect candidate points and the total number of black defect candidate points corresponds to each. 6. The defect inspection method according to claim 5, wherein if the value exceeds a predetermined value, it is determined that a defective portion exists.