JPH0882604A - Method for inspecting surface defect of steel plate - Google Patents

Abstract

PURPOSE: To detect a surface defect of a steel plate highly accurately by extracting multi-dimensional characteristic amounts such as a width of the surface defect, an area of the defective part, a lengthwise/breadthwise ratio, a maximum reflection intensity, etc., from a still image, thereby judging the defect. CONSTITUTION: A reflecting light of a light projected onto a steel plate 1 from a light source 13 such as a halogen lamp or the like is caught as a regular reflection light and an irregular reflection light. The image data is processed by averaging and averaging in XY directions, and also treated to obtain an average of a movement in a widthwise direction, and a threshold value, following a change of the surface of the steel plate in a longitudinal direction, whereby as many as about 40 kinds of multi-dimensional characteristic amounts are extracted. Among the characteristic amounts, particularly, lengths in XY directions, an area, a lengthwise/breadthwise ratio, a moment, the center of gravity and the like as size data, and a maximum intensity, a minimum intensity and an average intensity as reflection intensity data are extracted. Then, flaws are classified. A characteristic amount per flaw is extracted, thereby to judge a defect. An alarm is generated and an image of the defect is monitored. A classification accuracy for flaws is thus improved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a steel plate surface defect inspection method for highly accurately inspecting surface defects of thin steel plates and the like.

[0002]

2. Description of the Related Art Conventionally, a thin steel sheet having a surface defect of a cold-rolled thin steel sheet is used for manufacturing an exterior member of a product in which unevenness in the surface of the coated sheet causes unevenness in the appearance of an automobile or the like. Is not suitable. For this reason, conventionally, discrimination between a good product and a defective product has been performed at the time of pre-shipment inspection after rolling. This discrimination was performed by an inspector shining light on the surface of the thin steel sheet and directly inspecting the pattern of the reflected light, or by inspecting for the presence of a surface defect using an optical surface defect inspection device. . These conventional methods utilize the optical principle that light reflected from a defect-free surface causes uniform light scattering, whereas light reflected from a defect surface causes uneven scattering. Is. However, these methods could not inspect extremely fine surface defects with sufficient sensitivity. Also, since it is necessary to keep the test material stationary,
The running condition of the steel sheet could not be inspected.

Therefore, a laser type surface flaw inspection method has been proposed. In this laser type surface flaw inspection method, there is a simple method of determining the grade of defects such as heavy defects, medium defects, and light defects from the reflected light of the laser beam applied to the surface of the material to be inspected. In addition, Japanese Patent Publication No. 6-2986
As disclosed in Japanese Patent Laid-Open No. 4 (1994), the material to be inspected is irradiated with irradiation light, the irradiated portion is imaged by a camera, captured as a still image, and a defect is determined based on this still image. The surface of the surface defect and the SN ratio, which is the ratio of the brightness level of the defective part to the brightness level of the normal part,
A surface defect inspection method for determining a defect grade from the detected area and SN ratio has been proposed.

[0004]

The above-mentioned conventional simple method has a problem that the judgment is different from the judgment by human eyes because the judgment is based on the ratio of the luminance level of the defective portion and the luminance level of the normal portion. is there. In addition, Japanese Patent Publication 6-29864
In the case of the publication, the flaw is estimated from the area of the surface defect, the SN ratio which is the ratio of the luminance level of the defective portion and the luminance level of the normal portion, and the type of the flaw is a line called oyster flaw or sliver. Although the flaw is a flaw having a very high degree of harmfulness, it has a small area because it has a thin shape. Therefore, if the grade is judged based on the area, an erroneous judgment often occurs. Furthermore, in the case of oil adhesion or white stains (due to cold rolling oil), it has a certain area and the intensity of diffuse reflection is large. Therefore, depending on the maximum intensity of the area and the brightness level, it may cause a harmful flaw. However, there is a problem in that over-detection often occurs.

[0005]

In order to solve the problems as described above, the inventors of the present invention have conducted extensive studies and as a result, by detecting multidimensional feature amounts corresponding to dimensional accuracy and strength information, It is an object of the present invention to provide a steel sheet surface defect inspection method capable of highly accurately determining a surface defect despite dirt and the like. The gist of the invention is to irradiate a steel plate surface with illumination light, capture an image of the steel plate surface with a camera, capture it as a still image, and in the inspection method for determining a steel plate surface defect based on the still image, Steel plate surface characterized by detecting four dimensions of the steel plate surface defect width, defect area, aspect ratio, and maximum strength captured by the image as main feature amounts, and determining defects from the detected main feature amount extraction There is a defect inspection method.

[0006]

The present invention will be described in detail below with reference to the drawings. FIG. 1 is a schematic diagram showing equipment for implementing the present invention. As shown in FIG. 1, after the steel plate 1 is temper-rolled by a temper rolling mill 2, a fixed roll 4 or a bridle roll 5 via a bridle roll 3 is provided with a sensor unit 6 including a light source for irradiating illumination light. To do. An image of the irradiation state of the steel plate at the sensor unit 6 is picked up by a camera, sent to the signal processing board 7, the analysis / tuning WS 8 and the data storage video 9, the flaw map 10 and the image display 11 are transmitted, and transmitted to the processing computer. It consists of a configuration. Reference numeral 12 is a coiler.

FIG. 2 is an explanatory diagram showing signal processing for flaw inspection according to the present invention. As shown in FIG. 2, the light emitted from the light source 13 such as a halogen lamp onto the steel plate 1 is regarded as regular reflection and irregular reflection, and the averaging process and the XY direction averaging process are performed following the formation change in the longitudinal direction. And width direction moving averaging processing are performed and threshold processing is performed. Then, 40 kinds of multidimensional feature amounts are extracted, and in particular, XY direction length, area, aspect ratio, moment as dimension information, After extracting the maximum intensity, minimum intensity, average intensity, etc. as intensity information such as the center of gravity, perform defect classification, perform pitch judgment, alarm and image monitoring, and extract feature quantities for each defect. It is intended to improve the classification accuracy of defects and the classification accuracy of defects.

FIG. 3 is a diagram showing an evaluation method of a defect detection rate. As shown in FIG. 3, the flaw data is sampled online, the signal level of the flaw signal S for each flaw type and score is detected, the signal strength X is calculated, and the average value of the signal strength X and Standard deviation σ and threshold value (X _{1 to} X ₃ )
And the possibility of detection is evaluated.
For example, if thresholds X ₁ = 12 (linear flaws) X ₂ = 23 (dot flaws) and X ₃ = 30 (planar flaws) X ₁ , X
_The detection rate is evaluated with the probability that ₂ , X ₃ <X. And X
It has been found that 100% detection is possible with _{1 to} X ₃ ≧ 2.5σ.

FIG. 4 is an explanatory view showing a flaw type classification method according to the present invention. In the case of a defect type, if there is a feature amount of a defect to be classified that shows a completely different distribution from other defects, it is possible to easily classify by using this feature amount. However, flaws that can be separated using only one feature amount are rare, and as shown in FIG. 4, the feature amount distribution of the flaw A to be separated overlaps the distribution of other flaws B. In particular, when the flaw type B is a harmless defect, if only the feature amount A is set, it is overlooked or overdetected. In this case, as shown in the figure, it is necessary to classify using two feature quantities.

First, by using the feature amount A, a part of the flaw type B is stratified under the condition 1, and then using the other feature amount B, the remaining flaw type B is stratified under the condition 2 by using the feature amount A. The operation of depositing only seed A is performed. That is, the defect type A and the defect type B are a
When it is assumed that there are overlaps in _{1 to} a ₂ , since it is desired to separate the defect species A, it is first judged by the condition 1 whether or not the characteristic amount A> a ₂ and the characteristic amount A> a.
If it is ₂ , there is no overlap, so it can be clearly separated from the defective species.
When the feature amount A> a ₂ is not satisfied, there is an overlap, and therefore it is determined according to the condition 2 whether the feature amount B> b. If applicable, the defect type B, and if not, the defect type A.

[0011]

EXAMPLES In the equipment line shown in FIG. 1, a halogen lamp light source was used to irradiate a steel sheet to classify the detected flaws into flaw types. In the classification of flaw types, as features with oil, a feature amount as shown in Table 1 was selected from the feature that it has a certain area and the intensity of diffuse reflection is large (shining), and the range was set. Table 2 shows the characteristics of white stains, such as having a certain area, increasing the intensity of irregular reflection light, and having a small amount of change in regular reflection compared to the amount of change in irregular reflection. The feature amount was selected and the range was set.

[0012]

[Table 1]

[0013]

[Table 2]

Further, by using the detected flaw as the first step, the feature amount is analyzed for each flaw type, the flaw type is classified for each flaw type, and the harmless defect is removed, whereby the over-detection can be prevented. . That is, in order to classify into annealing stains, winding deviations, and white stains, if the surface is flat in step 1, the condition 1
Depending on the intensity and shape of diffused reflection light, it is separated into annealing stains or winding deviations, and it is separated into winding deviations or white stains according to the condition 2 specular reflection area / diffuse reflection area area ratio and aspect ratio. The winding deviation and the annealing stain are separated according to the light intensity, the shape and the like. Further, in step 4, grade classification is performed, and it is possible to separate the size of the winding misalignment and the size of the stain.

[0015]

As described above, by detecting the multidimensional feature amount corresponding to the dimensional accuracy and strength information according to the present invention, highly accurate surface defects can be determined despite oil stains, white stains and the like. This makes it possible to rationalize the adjusting process and rationalize personnel by automating with high precision, and it has an extremely excellent industrial effect.

[Brief description of drawings]

FIG. 1 is a schematic view showing equipment for carrying out the present invention,

FIG. 2 is an explanatory view showing signal processing for flaw inspection according to the present invention,

FIG. 3 is a diagram showing an evaluation method of a defect detection rate,

FIG. 4 is an explanatory diagram showing a flaw type classification method according to the present invention.

[Explanation of symbols] 1 steel plate 2 temper rolling mill 3 bridle roll 4 fixed roll 5 bridle roll 6 sensor part 7 signal processing board 8 analysis / tuning WS 9 data storage video 10 flaw map 11 image display 12 coiler 13 light source

Claims (1)

[Claims] 1. A steel plate surface is illuminated with illumination light, the steel plate surface is imaged by a camera, and is captured as a still image. In an inspection method for determining a steel plate surface defect based on the still image, the still image is captured. Steel plate surface defect width, defect area,
A steel plate surface defect inspection method, characterized in that four dimensions of aspect ratio and maximum strength are detected as main feature amounts, and defects are determined from the detected main feature amount extraction.