JPH03100882A - Defect inspection method by visual inspection - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention]

rIndustrial Application Fields] The present invention is a method for detecting defects by visual inspection, in which the area of the area to be inspected is determined based on an image taken of an area including the area to be inspected, and the product is determined to be non-defective when this area is within a predetermined range. This relates to testing methods.

[Conventional technology]

Conventionally, as a method for determining the area of a region to be inspected, a spatial region including the region to be inspected is imaged by an image input means such as a TV camera, and the region to be inspected is surrounded in the original image output from the image input means. By setting a judgment area of an appropriate shape and binarizing the density of pixels within the judgment area using a fixed or floating threshold set within this judgment area, the inside of the inspection area is divided into the area to be inspected. A method is known in which the area of the inspected area is determined after dividing the area outside the inspected area (for example, see Japanese Patent Laid-Open No. 63-44282).

[Problem to be solved by the invention]

In the above conventional method, when there are areas close to the inspection area that have concentrations in the same range as the inspection area with respect to the threshold value, the judgment area is attached to those areas. In this case, it becomes impossible to accurately determine the area of the region to be inspected. Therefore, if the position of the determination area is not set accurately enough, there is a problem in that the accuracy of defect detection decreases. The present invention aims to solve the above-mentioned problems, and it is possible to accurately determine the area without setting a judgment area for the inspected area, and to improve the appearance of the area with high defect detection accuracy. The present invention attempts to provide a method for detecting defects through inspection.

[Means to solve the problem]

In order to achieve the above object, the present invention captures an image of a spatial region including a region to be inspected by an image input means, and then calculates the area of the region 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 using visual inspection, in which the product is determined to be non-defective when the calculated area is within a predetermined range, multiple search starting points are set in parts that can be considered to be inside the inspected area, and each search starting point is Distinguishes between pixels within the inspection area and pixels outside the inspection area based on the magnitude relationship between the density of each adjacent pixel and a predetermined threshold, and sets a flag on the pixels determined to be within the inspection area. After that, the flagged pixel is set as the next search starting point, and all pixels adjacent to the search starting point for which no flag is set are determined to be outside the inspection area, and there are no other search starting points. If no pixels remain, the total number of flagged pixels is taken as the area of the region to be inspected.

[Effect]

According to the above method, a plurality of search start points are set in a region that can be considered to be inside the inspection area, and the number of search start points is determined based on the magnitude relationship between the density of each pixel adjacent to each search start point and a predetermined threshold. After identifying pixels within the inspection area and pixels outside the inspection area and setting a flag on the pixels determined to be within the inspection area,
The flagged pixel is set as the next search starting point, and all pixels adjacent to the search starting point for which no flag is set are determined to be outside the inspection area, and there are no other search starting points left. If not, the total number of flagged pixels is taken as the area of the area to be inspected, so multiple search starting points set in an area approximately approximate to the area to be inspected are used as seeds to spread around the search starting point. After forming a group of flagged pixels and filling the inspection area with the flagged pixels, the area of the inspection area is determined based on the number of pixels, and the area outside the inspection area is determined. Accurate area measurement can be performed without setting the area. As a result, it is possible to accurately determine the quality of the area to be inspected.

【Example】

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 TV camera, and the density of each pixel is converted into a digital signal in an analog-digital converter 2, and then converted into an analog-digital signal. The outline of the object to be inspected is extracted based on the original image obtained as the output of the converter 2. The original image is stored in the original image memory 3, and the arithmetic processing unit 4
Accordingly, a predetermined process described below is performed. That is, as shown in FIG. 2, the output of the image input device 1 illuminates the object O with a light source placed at a predetermined position, and images the spatial region including the object 0 with the image input bag W11. The original image obtained as shown in FIG. 3 is a grayscale image as shown in FIG. 3, in which areas having different densities are formed. The outermost square in FIG. 3 is the outline of the object to be inspected 0, and the circles, triangles, and squares shown in the object to be inspected O each indicate regions with different concentrations. Here, the outer shape and dimensions of the object to be inspected O are known, and
It is assumed that the approximate position of the object to be inspected 0 on the screen is also known. The triangular area is the inspected area M of interest, and its approximate position and shape are known, and as shown in FIG. , multiple search starting points P + , P 2 ,...
..., Po is set. Each search starting point P+, P2
．． . . . The following processing is sequentially performed on Ph. Here, it is assumed that the inspection area M has higher pixel density (that is, brighter) than the surrounding area. First, the search starting point p+ (coordinates are (X, yJ)) is As shown in the figure, the density fl(X1+h+3/J
0. ) is compared with a predetermined threshold SL. Here, n
, m=-1, 0, 1, except for the case where n=o and m=o, for the threshold SL, f + (X t+a+3'J+j> S L ,
A pixel for which -(1) is satisfied is considered to be a pixel of the region M to be inspected, a flag is set, and the pixel is set as a new search start point giving priority to the search start point P2. That is, the threshold value SL is set to a value that distinguishes the region M to be inspected from other regions. Furthermore, as described above, for the search starting point P1, after comparing the density of neighboring pixels (eight neighboring pixels) with the threshold value SL, the density of neighboring pixels for the flagged pixel is set to the threshold value SL. It is compared with SL. Here, pixels for which flags have already been set are not compared with the threshold value SL. As described above, if the pixel for which the flag has been set is used as the starting point for the next search and the comparison with the threshold value SL is performed one after another, the pixels that are considered to be in the area to be inspected M can be A group will be formed. In this way,
In the process of forming a group of pixels in the region to be inspected M, the initially set search starting points P,,P! ,...
...Flags are also set for some of the pls. The process of comparing the density of a pixel near the search starting point with the threshold value SL and identifying whether the pixel is in the inspection area M is performed by the following two steps.
The process continues until two stopping conditions are met. (2) Regarding pixels near the search start point, there is no pixel for which condition (1) is satisfied. ■All search starting point PIs set at the beginning. P2+..., a flag is set for Ph. If the above processing is performed, it is assumed that, for example, as shown in FIG. 6, dust D is present in the area to be inspected M, and the concentration of the part corresponding to the dust D is lower than that of the surrounding area. In this case, when the above process is performed for the shaded area in the inspection area M and flags are set for all pixels in the shaded area, the above-mentioned condition (2) is satisfied. However, the initially set search starting point remains in the area that is not shaded in the inspection area M, and the condition (■) is not satisfied, so the above process is continued, and as a result, Flags are set for all pixels even in areas that are not shaded.Flags are set for all pixels in the inspection area M, excluding the area corresponding to the dust D, so that the flag is set to 0 or more. become. As described above, after the stop condition ■■ is satisfied and the above processing is completed, the total number of pixels for which flags are set can be found on the surface 8IA of the inspection area M excluding defects such as dust D.
can be found. Therefore, the area A of the region M to be inspected is within the range (lower limit SL
It is sufficient to set the upper limit value SL, ) and determine that the product is non-defective when the following condition holds: SLL<A<SLd...(It). Furthermore, if the condition (2) is not satisfied, it is determined that a defect exists outside the inspection area. In the above embodiment, the area to be inspected M is brighter than the surrounding area, and the defect is darker than the area to be inspected. However, if the brightness and darkness are reversed, the size relationship when determining the area to be inspected M and defects is You can do it the other way around. Further, the threshold value SL, the lower limit value SLL for non-defective products, and the upper limit value SL are set to appropriate values depending on the contrast between the region to be inspected M and the surroundings and the contrast between the region to be inspected M and defects.

【Effect of the invention】

As described above, the present invention sets a plurality of search starting points in a region that can be considered to be inside the inspection area, and based on the magnitude relationship between the density of each pixel adjacent to each search starting point and a predetermined threshold value. After identifying pixels within the inspection area and pixels outside the inspection area and setting a flag on the pixels determined to be within the inspection area, the flagged pixel is used as the next search starting point and the search is performed. If all pixels adjacent to the starting point for which flags are not set are determined to be outside the inspection area, and there are no other search start points remaining, the total number of pixels with flags set is added to the inspection area. Since the area is set to , a group of pixels with flags set around the search start point is formed around the search start point using multiple search start points set in an area approximately approximate to the area to be inspected, and the flag is set. After the detected pixels fill the area to be inspected, the area of the area to be inspected is determined based on the number of pixels, and accurate area measurement can be performed without setting a judgment area outside the area to be inspected. . As a result, it has the advantage of being able to accurately determine the quality of the area to be inspected.

[Brief explanation of drawings]

Fig. 1 is a block diagram of a processing circuit corresponding to the method of the embodiment of the present invention, Fig. 2 is a perspective view showing the arrangement relationship between the inspected object and the image input device in the same as above, and Fig. 3 is an original image in the same as above. FIG. 4 is an explanatory diagram showing the relationship between the inspection area and the search starting point in the same example.
The figure is an operation explanatory diagram showing the vicinity of the search starting point in the same as above.
FIG. 6 is an explanatory diagram of the operation when there is dust in the inspection area in the same as above. 1 is an image input device, M is an area to be inspected, 0 is an object to be inspected 1, P2.・・・・・・Po is the starting point of the search.

Claims (1)

[Claims] (1) After capturing an image of a spatial region including the region to be inspected by the image input means, the area of the region to be inspected is determined based on the density of each pixel of the original image obtained by the image input means, and the determined area is within a predetermined range. In a defect inspection method using visual inspection, in which the product is determined to be non-defective when the area is within After identifying pixels within the inspection area and pixels outside the inspection area based on the size relationship with the threshold value and setting a flag on the pixels determined to be within the inspection area, the flagged pixels is set as the next search starting point, and if all pixels adjacent to the search starting point for which no flag is set are determined to be outside the inspection area, and there are no other search starting points remaining, the flag is set. 1. A defect inspection method by visual inspection, characterized in that the total number of pixels detected is the area of an area to be inspected.