JPH0663984B2 - Defect detection method - Google Patents

Description

TECHNICAL FIELD The present invention relates to a defect detection method used, for example, in detecting microscopic defects on the surface of a steel sheet.

[Conventional technology]

As a method for detecting a microscopic defect on the surface of a steel sheet, a method based on the surface image is known, and in such a detection method, since the defect is often a brightness change portion, the surface image Defects are often detected on the basis of an enhanced image such as a differential image obtained by differentiating.

[Problems to be Solved by the Invention]

However, when a defect is detected by listening to the differential image in which the surface image is emphasized, when there is a rapid change in the background such as pickling unevenness, this background change is also emphasized. There is a problem that it is difficult to detect such defects.

The present invention has been made in view of such circumstances, to obtain an emphasized image by the addition process or the absolute difference process of the absolute value of the differential image and another differential image obtained by shifting the differential image by a predetermined amount in the differential direction, An object of the present invention is to provide a defect detection method capable of accurately and easily detecting a minute defect even if there is a change in the background by detecting a defect based on this emphasized image.

[Means for Solving the Problems]

A defect detection method according to the present invention is a defect detection method for obtaining a surface image of a detected object, image processing the surface image to detect surface defects of the detected object, and differentially processing the surface image in one direction. To obtain a first differential image, and to shift the first differential image in the direction by a predetermined amount to obtain a second differential image,
The addition of the absolute value of the first differential image and the absolute value of the second differential image, or the absolute value of the difference between the first differential image and the second differential image, The enhanced image of the surface image is obtained, the enhanced image is binarized, and then the small particles are removed or the image is reduced in a direction perpendicular to the direction to detect the surface defect of the detected object.

[Action]

According to the defect detecting method of the present invention, first, the surface image of the object to be detected is differentiated to obtain the first differential image. The first differential image is shifted in the differential direction to obtain the second differential image. An enhanced image is obtained by performing an addition process of the absolute values of the two types or an absolute value process of the difference between the two types of differential images. Next, after the binarization process is performed on the emphasized image, a small particle removal process or a reduction process in a direction perpendicular to the differential direction is performed to detect a defect.

In the obtained emphasized image, the change in the minute area is further emphasized and the change in the wide area is not emphasized, so that the minute defect can be accurately detected without being affected by the change in the background.

〔Example〕

Hereinafter, the present invention will be specifically described with reference to the drawings illustrating the embodiments. In the present embodiment, a case will be described in which microscopic defects on the surface of the steel sheet after the pickling treatment are detected.

FIG. 1 is a schematic diagram showing an implementation state of the defect detection method according to the present invention, in which 1 denotes a steel plate as an object to be detected for a microscopic defect on the surface. After the steel sheet 1 is subjected to pickling treatment, it is conveyed in the front and back direction of FIG. A strobe 2 and a camera 3 are provided above the steel plate 1, and a surface image of the steel plate 1 can be obtained on the camera 3 by illumination from the strobe 2. The camera 3 is connected to an image processing device 4 that performs image processing on the surface image obtained by the camera 3 to detect minute defects.

Next, the operation will be described.

The steel plate 1 conveyed below the camera 3 is imaged by the camera 3 by the illumination from the strobe 2, and a surface image thereof is obtained. The surface image obtained by the camera 3 is input to the image processing device 3 and subjected to image processing in the procedure described below to detect a micro defect.

The pickling unevenness is observed on the surface of the steel sheet after the pickling treatment, and in the obtained surface image, the luminance level of the portion where the minute defect exists and the pickling unevenness is high. FIG. 2 is a schematic diagram showing the brightness level of the surface image, which includes minute defect portions (FIG. 2 (a)) and uneven pickling portions (FIG. 2 (b)).
The brightness level of the and part is high. However, the area where the minute defects are present has a narrower brightness change area than the area where the pickling is uneven.

Such a surface image is differentiated in the left-right direction in FIG. 2 to obtain a first differential image. A shown by a solid line in FIG. 3 is a schematic diagram showing the brightness level of the differential image obtained by differentiating the surface image. The part (a) in the figure corresponds to the minute defect part, and the part (b) in FIG. The part corresponds to the uneven pickling part. Since the minute defect portion has a narrow brightness change area, the positive and negative amplitude distributions are close to each other at the brightness level of the differential image, and the uneven brightness area has a wide brightness change area, so the brightness level of the differential image is separated. It exhibits positive and negative amplitude distributions.

Then, the first differential image is shifted in the differential direction by a predetermined amount to obtain the second differential image. B shown by a broken line in FIG.
It is a schematic diagram which shows the brightness level in this 2nd differential image, and shows the waveform which shifted the waveform of A to the right direction of FIG.

Next, the absolute value of the luminance level of the first differential image and the second
Processing for adding the absolute value of the brightness level of the differential image of (absolute value addition processing), or processing for obtaining the absolute value of the difference between the brightness level of the first differential image and the brightness level of the second differential image ( An absolute image is obtained by performing a difference absolute value process). The absolute value addition processing and the difference absolute value processing are specifically shown by the following equations (1) and (2). However, n
Indicates a predetermined amount shifted in the differential direction.

g (x, y) = | f (x, y) | + | f (x + n, y) | ... (1) g (x, y) = | f (x, y) -f (x + n, y) | (2) FIG. 4 is a schematic diagram showing the brightness level in the emphasized image obtained by performing these processes. As can be seen from FIG. 4, the luminance level is increased by being emphasized in the minute defect portion (a), but is not emphasized in the uneven pickling portion (b) and the luminance level does not change.

Next, after performing binarization processing at an appropriate level on the emphasized image exhibiting such a brightness level, small particles are removed or reduction processing is performed in a direction perpendicular to the differentiation direction to remove noise components, Detect shrinkage defects.

Therefore, in the detection method of the present invention, the pickling unevenness portion is not emphasized but the reduction defect portion is emphasized, so that the pickling unevenness can be accurately detected without erroneously detecting the pickling unevenness as a minute defect. it can. In other words, without the effect of uneven pickling,
Minute defects can be detected.

Although the apparatus configuration shown in FIG. 1 is configured to detect minute defects on the upper surface of the steel plate 1, when another strobe and camera are provided below the steel plate 1 to perform similar image processing. Of course, it is of course possible to detect minute defects on both sides at the same time.

〔The invention's effect〕

In the defect detection method of the present invention, since an enhanced image is obtained by performing absolute value addition processing or differential absolute value processing on two types of differential images that are shifted in the differential direction, when there is a background change such as pickling unevenness. Even if there is, the minute defect can be accurately detected.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing an implementation state of a defect detecting method according to the present invention, FIG. 2 is a schematic diagram showing a luminance level of a surface image, FIG. 3 is a schematic diagram showing a luminance level of a differential image, and FIG. FIG. 6 is a schematic diagram showing the brightness level of an emphasized image. 1 ... Steel plate, 2 ... Strobe, 3 ... Camera, 4 ... Image processing device

Claims (1)

[Claims] 1. A defect detection method for obtaining a surface image of an object to be detected and image-processing the surface image to detect surface defects of the object to be detected, wherein the surface image is differentially processed in one direction. Differential image is obtained, the first differential image is shifted in the direction by a predetermined amount to obtain a second differential image, and the absolute value of the first differential image and the absolute value of the second differential image are added. Processing, or the first
Of the surface image is obtained by performing absolute value processing of the difference between the differential image and the second differential image, and the enhanced image is binarized, and then the small particles are removed or the direction perpendicular to the direction is obtained. A defect detection method, characterized in that a surface defect is detected by performing a reduction process in a direction.