JPS62235550A - Defect inspection - Google Patents

Abstract

PURPOSE:To achieve an automatic and higher-speed inspection, by performing a contraction processing and an expansion processing of a binary coded image sequentially to generate a corrected image allowing vanishment of a pseudo normal part as existing in a defect part of an original image. CONSTITUTION:An image 2 shown in a drawing (a) is subjected to a contraction processing to contact the image 2 as a whole. This contracts a normal part 4 as shown by (b) from the original size and also does a pseudo normal part 5. Then, this contraction processing is repeated several times until the pseudo normal part 5 vanishes completely as shown by (c). The frequency may be set beforehand by being allowed for the size of an assumed false normal part 5. Then, an expansion processing is conducted as many time as the contraction processing is done to enlarge the image 2 to the original size as a whole as shown by (d) and (e). With such an arrangement, a corrected image 6 with the pseudo normal part 5 alone removed from the original image 2 is obtained without the restoration of the pseudo normal part 5 vanished in the contraction processing process while the original normal part 4 is returned to the original size.

Description

DETAILED DESCRIPTION OF THE INVENTION "Field of Industrial Application" The present invention relates to a method for inspecting defects in, for example, nuclear fuel pellets.

``Conventional technology 1'' Conventionally, this type of inspection was performed using a V& image of the outer surface of the object to be inspected.
This is often done mechanically by creating a digitized image and processing that image.

For example, nuclear fuel pellets as shown in Figure 2! When inspecting the outer peripheral surface of a nuclear fuel pellet 1, first, as shown in Figure 3 (a), a photographic light is applied to the outer peripheral surface of the nuclear fuel pellet 1 and the reflected light is received, and the intensity of the reflected light from each part is measured. The area is binarized using a predetermined threshold value to create a binarized expanded image 2 of the entire circumferential surface as shown in FIG. 3 (b). At this time, if there is a defective part 1a such as a chip on the outer surface of the pellet 1, the photographing light is diffusely reflected at the defective part 1a and the normal part 1b
Since the brightness is different from that, in the created image 2, the defective part 1a of the pellet l will be shown as the defective part 3, and the normal part 1b of the pellet l will be shown as the normal part 4.

Then, from this image 2, the defective part 3 or the normal part 4 is
The size of the pellet is measured using a preset image processing program, and the measurement results are compared against criteria to mechanically determine whether or not the pellet L is acceptable. That is, if the size of the defective portion 3 is greater than or equal to a predetermined value, or if the size of the normal portion 4 is less than or equal to a predetermined value, the pellet 1 is determined to be defective.

"Problems to be Solved by the Invention" However, in the conventional method described above, accurate determination may not be possible depending on the shape of the defective portion 1a of the pellet l.

For example, if there is a part in the defective part la that is almost parallel to the normal part 1b, diffuse reflection will not occur in that part as shown in Figure 4 (a), and it will be normal like the normal part 1b. may be reflected. In this case, the image of this part has the same brightness as the normal part 4, and as shown in FIG. (a portion that appears as an island) appears in the defective portion 3 in the form of an island.

In such cases, conventional general image processing programs
It is not possible to distinguish between the above-mentioned pseudo-normal area 5 and the original normal area 4, and such pseudo-normal area 5 is also classified as normal area 4.
I had decided that.

As a result, a defective pellet may be incorrectly determined to be a good product.

For this reason, in the past, in order to avoid such erroneous determinations, checks had to be performed by inspectors, making it difficult to automate inspections and improve efficiency.

The present invention has been made in view of the above circumstances, and it is an object of the present invention to provide an inspection method that can exclude the influence of pseudo-normal parts and make accurate determinations.

"Means for Solving the Problems" This invention images the outer surface of an object to be inspected to create a binarized image, and calculates the size of a defective part or a normal part appearing in the binarized image. When inspecting the size of defects such as chips on the outer surface of the object to be inspected by measuring, by sequentially performing contraction processing and expansion processing on the binarized image, defects in the original image can be detected. This method is characterized by creating a corrected image in which the pseudo-normal parts that existed in the part disappear, and then measuring the size of the defective part or the normal part in the corrected image.

"Example" Hereinafter, an example of the present invention will be described with reference to FIG.

Figures 1 (a) to (e) show that the original normal area is different from the image 2 in which the pseudo-normal area 5 exists as an island in the defect area 3, as shown in Figure 4 (b). The image processing procedure for measuring only the size of 4 or the size of the entire original defective part 3 excluding the pseudo-normal part 5 is shown in a schematic % formula % First, for image 2 shown in FIG. Then, the entire image 2 is shrunk by performing a shrinking process. ° This results in (
As shown in b), the normal part 4 shrinks from its original size (indicated by the broken line in the figure), and the pseudo-normal part 5 also shrinks in the same way. Then, as shown in (c), the contraction process is continued a predetermined number of times until the pseudo-normal part 5 completely disappears. Note that the number of times this contraction process is performed may be set in advance in consideration of the expected size of the pseudo-normal part 5.

Then, as shown in (d) and (e), expansion processing is performed as many times as the contraction processing performed above to return the entire image 2 to its original size. As a result, although the original normal part 4 returns to its original size, the pseudo-normal part 5 that disappeared during the shrinkage process described above will no longer be restored, and only the pseudo-normal part 5 will be restored from the original image 2. A corrected image 6 is obtained in which .

And this revised image @! 6, the size of the defective part 3 or the normal part 4 is measured in the same manner as in the past, and the result of the measurement is compared with the criteria to judge whether the pellet is acceptable or not.

According to the above procedure, since the pseudo-normal area 5 has disappeared prior to measurement, the pseudo-normal area 5 is replaced with the normal area 4 as in the conventional method.
It is possible to accurately determine pass/fail by measuring only the original defective part 3 or normal part 4 without making any misunderstanding. Therefore, the reliability of the inspection is improved, and the check by an inspector, which was necessary in the past, can be omitted.

The present invention is applicable not only to the inspection of the circumferential surface of nuclear fuel pellets as described above, but also to the inspection of the end faces thereof, and further to the inspection of general materials other than nuclear fuel pellets by image processing. Of course, it can also be applied in the same way.

"Effects of the Invention" As explained in detail above, according to the present invention, a corrected image is created in which the pseudo-normal areas disappear by shrinking the image and then expanding it again, and from this corrected image, the defective areas are Alternatively, since the normal part is measured, it is possible to accurately determine pass/fail, and it is also possible to automate and speed up the test.

[Brief explanation of drawings]

FIGS. 1A to 1E are diagrams schematically showing the procedure of an inspection method according to an embodiment of the present invention. FIG. 2 is a perspective view showing the shape of nuclear fuel pellets. Figure 3 (a) is a cross-sectional view showing the imaging state of the outer peripheral surface of a nuclear fuel pellet, Figure 3 (b) is a diagram showing a binarized image, and Figure 4 (a) shows a pseudo-normal area. FIG. 4(b) is a cross-sectional view showing the imaging state in this case, and is a diagram showing an image in which a pseudo-normal area appears. ■・・・・・・Pellet for nuclear fuel (object to be inspected), la・
...Defect part, 2...Binarized image, 3...
...Defect part, 4...Normal part, 5...
・Pseudo-normal part, 6... Corrected image.

Claims (1)

[Claims] By capturing an image of the outer surface of the object to be inspected and creating a binarized image, and measuring the size of the defective part or normal area appearing in the binarized image, chips etc. on the outer surface of the object to be inspected can be detected. When inspecting the size of the defective area, the pseudo-normal area that existed in the defective area of the original image was eliminated by sequentially performing contraction processing and expansion processing on the binarized image. A defect inspection method characterized by creating a corrected image and measuring the size of a defective part or a normal part in the corrected image.