JPH0410666B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field The present invention relates to a method for detecting defects such as cracks and chips on the surface of an object to be detected, based on image data obtained by imaging the object.

BACKGROUND ART There is a method of detecting shape defects such as cracks and chips on the surface of a detection object by image processing.
In the conventional technology, the detected object is imaged with a television camera, the image signal from the television camera is digitized, the digitized image data is stored in an image memory, and the stored image data is processed to detect defects in the detected object. is being detected.

Image processing for detecting defects requires pattern recognition of grayscale images. A line drawing is created using edges, which are the contours of the detected object, from an original image obtained by capturing an image of the detected object, and the features of the line drawing are extracted to perform pattern recognition. To extract edges from a grayscale image, a differential method is generally used because it is sufficient to extract changes in density within the image.

FIG. 7 is for explaining the differential processing. FIG. 7(1) shows arbitrary 3×3 pixels on the original image. Assuming that the reference marks A to I indicate the density at each pixel, the vertical differential value ΔV of the pixel with density E is expressed by the following first equation, and the horizontal differential value ΔH is expressed by the following equation 1. It is shown by equation 2.

ΔV=(G+H+I)−(A+B+C)…(1) ΔH=(C+F+I)−(A+D+G)…(2) Then, the differential value |e| of the pixel of density E is expressed by the following third equation.

|e|=√() ² +() ² …(3) Also, the differential direction ∠e at that pixel E is the following fourth
It is expressed as the formula.

∠e=tan−1ΔV/ΔH+π/2 (4) When such an operation is performed for all pixels of the original image, a differential image and a differential direction in the image are obtained.

Next, edge extraction processing will be explained. Figure 7
(2) extracts arbitrary 3×3 pixels in the differential image and indicates each pixel by its differential value |e| to |i|. Focusing on a pixel having a differential value |e|, the differential value of the pixel is compared with two differential values located at 90 degrees with respect to the differential direction ∠e.
Only when the differential value |e| is larger than the two neighboring differential values, an edge flag is set at the address corresponding to the pixel on the edge extracted image. For example, |e| is the threshold value of the differential value |e|, and |e
If | ₂ is a differential value in a direction perpendicular to the differential direction of the pixel, an edge flag is set if the following equation 5 is satisfied.

|e| ₁ <|e|> |e| ₂ (5) By performing this operation for all pixels, an edge line extracted image is obtained.

From the edge line extraction image thus obtained, the amount of deformation of the edge line is determined in order to determine the state of the defect. In the prior art, the angular difference between the edge flag and the next adjacent edge flag's column address is determined based on the row and column address of the edge flag, and the degree of deformation is determined based on the magnitude of the angle. With such a method, even if it is possible to detect the degree of deformation of the edge line, it is difficult to detect the magnitude of the change, which indicates the area of the defect.

Purpose An object of the present invention is to provide a defect detection method that can detect defects such as cracks and chips occurring on the surface of a detection object at high speed and with high accuracy.

Embodiment FIG. 1 is a block diagram showing the configuration of an image processing apparatus capable of implementing the present invention. The object to be detected is imaged by an industrial television camera 1. An analog signal that is image data of an object to be detected is output from an industrial television camera 1 to an analog/digital (A/D) converter 2. The A/D converter 2 digitally converts the analog signal from the industrial television camera 1 and supplies the converted signal to the differentiating circuit 3. In this differentiation circuit 3, the calculation processing shown in the first to fourth equations described in the background art is performed, and the differential value |e| of the pixel corresponding to the gray level difference of the edge of the image and the differential direction of the pixel ∠ e is required. This process is performed for all edge lines pixel by pixel. The differential value |e| obtained by the differentiator circuit 3 and its differential direction ∠e
is stored in the memory 4 and output to the edge line extraction circuit 5. The edge line extraction circuit 5 performs arithmetic processing as shown in equation 5 based on the differential value |e| computed by the differentiator circuit 3 and its differential direction ∠e. Edge flag ef obtained by calculation processing in edge line extraction circuit 5
is output to the memory 4 and stored. In the edge line tracing circuit 6, the threshold value |e in the fifth equation
For the edge line erased by | ₁ , the necessary edge line is regenerated again by a specific evaluation function calculation from the edge of the edge line that remains above the threshold value |e|, and its value is stored in the memory 4. Output. Based on the data as described above, edge feature extraction processing is performed in the image determination unit 7. The image judgment unit 7 uses the original image data Z and the differential value |e
｜, a swap memory 8 that stores each data of the differential direction ∠e and the reproduced edge line, and a CPU (central processing unit) 9 that performs feature extraction processing based on the edge flag using the stored data and determines its quality. Become. A control circuit 10 is provided for setting the threshold value and controlling the entire image processing apparatus.

The defect detection method of the present invention will be explained with reference to FIGS. 2 to 4. FIG. 4 is a block diagram showing a defect detection method according to an embodiment of the present invention. 1st
When the image of the object to be detected is converted into an edge by the image processing apparatus shown in the figure, it becomes as shown in FIG. 2(1). In FIG. 2(1), the reference mark L indicates the edge line of the detected object for which an edge flag has been set, and the arrow indicates its differential direction ∠e. Further, a changing portion of the edge line indicated by reference numeral K is a defective portion of the object to be detected. First, the edge flag es for starting scanning and its differential direction data are extracted. Next, an address is added by shifting m pixels in a direction perpendicular to the inside of the detected object with respect to the direction of the edge flag. Then, as shown in FIG. 2(2), a scanning line L1 is obtained in which addresses are added by shifting n pixels in the direction of the edge flag. Further, as shown in FIG. 2(3), a scanning line L2 is obtained by shifting n pixels in the direction opposite to the direction of the edge flag and adding addresses. When the aforementioned addresses are combined, a scanning line L3 shown in FIG. 2(4) is obtained. FIG. 3 is an enlarged view of the section shown in FIG. 2(4). If the intersections of the edge line L and the scanning line L3 are P0 and P1, then based on these intersections P0 and P1,
The following calculations are performed. For the intersection point P0, first, a point-symmetric address is found for each pixel in the direction of the scanning line L3, and the gray value Z of that address is calculated.
is extracted. Then, the following calculations of the sixth and seventh equations are performed.

ΔZ ₁ =Za ₁ −Zb ₁ ΔZ ₂ =Za ₂ −Zb ₂ 〓 ΔZn＝Nan−Zbn …(6) ΣΔZn＝ΔZ ₁ +ΔZ ₂ +…+ΔZn …(7) In the above formula, Za ₁ to Zan are each pixel The gradation values at a ₁ to an are shown, Zb ₁ to Zbn are the gradation values at each pixel b ₁ to bn, and ΔZ ₁ to ΔZn are the gradation differences.

Each shade difference ΔZ ₁ ~
ΔZn and its total sum ΣΔZn are calculated. Next, the comparison circuit compares the sum ΣΔZn with a predetermined threshold, and when it is above the threshold, the object to be detected is a defective product with a defect, and when it is below it, it is a good product. Judgment is made.

FIG. 5 is a diagram for explaining another embodiment of the present invention, and FIG. 6 is a block diagram showing a defect detection method according to another embodiment of the present invention. In this embodiment, an operation for detecting a defective portion K generated at a corner of an object to be detected as shown in FIG. 5(1) will be described. First, an edge line E indicating the outline of the object to be detected is determined as in the previous embodiment. Figure 5 (2) is
FIG. 5 is an enlarged view of the section in FIG. 5(1). As shown in Figure 2 (2), defective part K of edge line E
Extend the straight line up to , and find the apex C where no defect occurs. Then, the address of the vertex C is determined, and a scanning line S is set inward of the object to be detected using the vertex C as the scanning starting point. At this time, the scanning line S coincides with a straight line that bisects the angle formed by the virtual extension straight lines E1 and E2 of the edge line E and the vertex C. Then, the vertex C is set as the scanning starting point,
Pixels on the scanning line S shifted by m pixels in each direction of the virtual extension straight lines E1 and E2 are scanned. Next, it is determined whether an edge flag exists on the scanning line S from the scanning starting point C. Figure 5(2)
When an edge flag exists at the intersection of the scanning line S and the edge line E of the object to be detected, which is indicated by reference numeral Q, the following processing is performed. Figure 5(3)
2 is a further enlarged view of FIG. As shown in FIG. 5(3), pixels a ₁ to symmetrical in the scanning line S direction with respect to the point indicated by reference mark Q are
The shading values Za ₁ -Zan, Zb ₁ -Zbn of an, b ₁ -bn are obtained. And its gradation value Za ₁ ~ Zan, Zb ₁ ~ Zbn
Based on this, the tone differences ΔZ ₁ to ΔZn of the original image as shown in the above-mentioned formula 6 are calculated. Perform such calculations and use the results of the calculations to calculate the seventh
The sum ΣΔZn is calculated as shown in the equation. Then, in the same manner as in the above-mentioned embodiment, the sum ΣΔZn is compared with a predetermined threshold, and when it is above the threshold, the object to be detected is considered to be a defective product with a defect, and when it is less than that, it is considered to be a good product. A judgment will be made.

In the above-described embodiment, the grayscale values of each pixel are compared with respect to the reference point Q, but the average grayscale values of areas of a plurality of pixels may be compared. In such a case, detection errors due to dust or the like can be prevented.

In the embodiments described above, the present invention is carried out based on gray pixels, but it can also be carried out based on image data of a color image.

Effects As described above, according to the present invention, an image of an object to be detected is captured, the detected contents are stored, a reference position for inspection is determined, and a pair of objects that are symmetrical with respect to a predetermined position along the reference position for inspection is set. Since the detected contents at the positions forming the position are compared with each other and the size of the defect is detected based on the comparison result, the defect can be detected at high speed and with high accuracy.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an image processing apparatus according to an embodiment of the present invention, and FIG.
The figure is a diagram for explaining a defect detection method according to an embodiment of the present invention, FIG. 3 is a diagram for explaining a detection method according to an embodiment of the present invention in relation to FIG. 2, and FIG. A block diagram showing a defect detection method according to one embodiment of the present invention, FIG. 5 is a diagram for explaining another embodiment of the present invention, and FIG. 6 shows a defect detection method according to another embodiment of the present invention. The block diagram in FIG. 7 is a diagram for explaining the image processing method. DESCRIPTION OF SYMBOLS 1...Industrial television camera, 2...Analog/digital converter, 3...Differential circuit, 4...Memory, 5...Edge line extraction circuit, 6...Edge line tracing circuit, 7...Image judgment unit, 8...Swap memory, 9...
CPU, 10...Control circuit, E, L...Edge line, L
1, L2, L3, S...scanning line.

Claims (1)

[Claims] 1. Image the detection area, store the detection contents in a predetermined detection mode such as shading, detect the boundary of the detection area, and space the area inward from the boundary by a predetermined displacement amount, Also, when a defect etc. is assumed, a reference position for inspection parallel to the boundary is determined, and the detected contents at a pair of positions symmetrical with respect to the predetermined position along this reference position for inspection are compared with each other. A defect detection method characterized by detecting the size of a defect based on a comparison result.