JP5354187B2 - Traveling material surface quality judging device and surface quality judging method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for determining a surface quality of a moving material, which can accurately determine the surface quality of the moving material. <P>SOLUTION: In the method, bright portions and dark portions are extracted from an image which is obtained by photographing a surface of the moving material by using respective thresholds of the bright portions and the dark portions being set beforehand, and the extracted bright and dark portions are classified into one or more bright defects and one or more dark defects by using a plurality of feature quantities containing a dimension feature quantity and a brightness feature quantity, and the surface quality of the material is determined on the basis of an occurrence density of the classified bright and dark defects. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a traveling material surface quality determination device and a surface quality determination method that inspect surface defects of a traveling material such as a steel strip and determine the surface quality thereof.

  In the production line of steel strips such as cold rolled steel sheets and surface-treated steel sheets, the surface is inspected for the presence of severe surface defects (hereinafter referred to as “severe defects”) such as baldness, scale wrinkles, and oyster wrinkles. It is extremely important for quality assurance to prevent leakage.

  On the other hand, on the surface of the steel sheet, there are cases where extremely mild surface defects that do not reach harmful levels and minor surface heterogeneous parts (hereinafter referred to as light defects) due to changes or abnormalities in production line operation conditions occur. is there. These minor defects are harmless on their own, but if they occur densely, they may indicate some abnormality in the operating state, and may develop into harmful defects, so they are treated early. It is important for quality control to discover and correct operating conditions.

  In the production line of steel strip, many surface defect inspection devices using an image pickup method have been installed so far and play an important role in ensuring surface quality. These conventional inspection apparatuses separate and extract the above-mentioned serious defect from the formation portion by binarizing or multi-leveling an image of the surface of the steel sheet taken by a camera with a predetermined threshold value. Since the light defect generally has a small image brightness difference with a healthy formation part, there is a problem that overdetection occurs when trying to detect it by the same processing as a heavy defect. Not subject to inspection.

  As a technique for detecting the above-mentioned light defects, Patent Document 1 discloses a surface quality identification method for evaluating light defects on the surface of a steel sheet according to the density. (1) First, discriminate defects into heavy and light defects. (2) For light defects, the number of steel strips is counted for each specified unit length L1 (5-20m), which exceeds the allowable value. (3) For each specified unit length L2 (500 to 1000m) of the steel strip, the number of heavy defects and the number of dense defects are counted, and this is the predetermined allowable number. (4) Count the number of heavy defects and dense defects over the entire length of the steel strip, and compare this with a predetermined allowable number range to identify the surface quality of the steel strip. Is the method. In this method, since light defects are processed based on the idea of considering only when they occur densely, they can be separated from ground noise with relatively low density, and overdetection is suppressed. There is a possibility that minor defects can be detected.

Japanese Patent Publication No.59-22894

  However, the technique disclosed in Patent Document 1 can grasp the tendency of the surface quality to some extent when light defects of the same type are densely generated over the entire width as shown in FIG. Although effective, there are a number of problems:

That is,
1) Light defects may occur over the entire width of the steel strip. In general, however, as shown in FIG. 6B, light defects are densely formed in a specific strip width direction position. For this reason, especially when the width of the band of the generated portion is small, the generated density is underestimated and the surface quality cannot be determined appropriately.
2) In many cases, the strip width of the steel strip itself varies from 1.5 to 2 times, so even if light defects occur with the same density over the entire width, the judgment results differ depending on the strip width ( The wider the steel plate, the harder the judgment result).
3) There are various types of light defects, but these cannot be determined separately. For example, as shown in FIG. 6 (c), when two or more types of light defects are mixed, different light defects are added together and counted as a dense light defect. For this reason, even if the density of each light defect is relatively small, these are combined and processed, resulting in over-determination.
4) There are various types of minor defects, both large and small, but since these minor defects are processed equally, the determination result of the surface quality does not match the actual condition.

  The present invention has been devised in view of these problems of the prior art, and it is an object of the present invention to provide a surface quality determination method for a traveling material that can accurately determine the surface quality of the traveling material.

  The invention according to claim 1 of the present invention is an image input means for inputting an image obtained by imaging a traveling material surface, a light / dark area extracting means for extracting a bright defect candidate area and a dark defect candidate area from the input image, Feature quantity calculation means for calculating the dimensional feature quantity and the brightness feature quantity of the extracted bright defect candidate area and dark defect candidate area, and each of the defect candidate areas using one or more bright defects and 1 A traveling material surface quality judgment device comprising: a defect classification means for classifying two or more dark defects; and a quality judgment means for judging the surface quality of the material from the occurrence density of the classified bright defects and dark defects. is there.

  Further, the invention according to claim 2 of the present invention is the traveling material surface quality determination device according to claim 1, wherein the quality determination means defines the severity in advance for each type of bright defect and dark defect by a weighting factor. Then, the material surface is divided into a plurality of rectangular small areas, and for each small area, the number of bright defects weighted defects and the number of dark defect weighted defects are determined, based on the number of weighted defects, A surface quality determination device for a traveling material, characterized by determining a defect grade for each divided small region.

  Further, the invention according to claim 3 of the present invention is the quality determination device for the running material according to claim 1 or 2, wherein the defect occurrence density is determined for a serious defect having a severity greater than or equal to a predetermined value. In addition to the determination based on this, a quality determination device for a traveling material is characterized in that the quality is determined for each individual defect.

  Further, the invention according to claim 4 of the present invention extracts a bright defect candidate region and a dark defect candidate region from a captured image of the surface of a traveling material by using a threshold value set in advance for each of a bright portion and a dark portion, The extracted bright defect candidate area and dark defect candidate area are classified into one or more bright defects and one or more dark defects using the feature quantity including the dimension feature quantity and the brightness feature quantity, and the classified bright defect and dark defect A surface quality determination method for a running material, characterized in that the surface quality of a material is determined from a defect generation density.

  In the present invention, when a light defect is determined, it is classified into a light defect and a dark defect according to the polarity of the image luminance, and both are individually evaluated. Therefore, different types of light defects are mixed and evaluated. This makes it possible to accurately determine the surface quality. In addition to light and dark polarities, light defects are classified according to the degree of image features, and the density is evaluated by multiplying the weighting factor according to the degree of defects. Judgment is possible.

  Furthermore, in the present invention, the material surface is divided into a plurality of rectangular small areas, and the grade of light defects is determined in each small area, so that the surface quality distribution on the material surface can be accurately grasped, and light defects are Even if the material occurs in a specific width direction or the width of the material fluctuates, the material surface quality can be determined based on the same reference regardless of the variation.

It is a figure which shows the structural example of the surface quality determination apparatus of the traveling material which concerns on this invention. It is a figure which shows the example of a process sequence of the surface quality determination method of the traveling material which concerns on this invention. It is a figure which shows the example of a process sequence in one Example of this invention. It is a figure which shows an example of the weighting coefficient which concerns on this invention. It is a figure which shows the evaluation threshold value for classifying the grade of the small area | region which concerns on this invention. It is a figure which illustrates the problem of a prior art typically.

  A mode for carrying out the present invention will be described below with reference to the drawings, taking as an example the case where a steel plate is a quality judgment target. FIG. 1 is a diagram illustrating a configuration example of a surface quality determination device 1 for a traveling material (steel plate) according to the present invention. In general, since the surface inspection apparatus is functionally divided into an imaging unit and a signal processing / defect determination unit, the surface quality determination apparatus 1 of the present invention corresponds to the function of the signal processing / defect determination unit, It may be considered as a surface inspection device in combination.

  The surface quality determination device 1 is input with an image input unit 4 that inputs image data obtained by imaging the surface of the inspection target material traveling from the imaging unit 2 (an imaging device such as a line sensor camera or an area sensor) of the surface inspection device. A data storage unit 5 that stores data such as image data and coefficients necessary for calculation; a calculation processing unit 6 that classifies defects from the image data and determines surface quality based on the occurrence density of defects after classification; A result output unit 7 is provided for displaying the results on a monitor, printing them on a printer, or transmitting them to an operating computer.

  The image input unit 4 also receives a line signal 3 such as PLG for capturing a signal with uniform resolution in the moving direction even when the traveling speed of the traveling material changes, and the signal is synchronized with the traveling amount of the traveling material. Is stored in the image data storage unit 51 of the data storage unit 5.

  The arithmetic processing unit 6 includes a light / dark region extraction unit 61, a feature amount calculation unit 62, a defect classification unit 63, a weighted number calculation unit 64 for each small region, a quality grade determination unit 65 for each small region, and a quality grade determination for the entire steel plate. Part 66. The light / dark region extraction unit 61 reads image data from the image data storage unit 51, and shows a bright portion and a dark portion with respect to the formation portion (sound portion) on the surface of the running material according to a preset threshold value, That is, a bright defect candidate area and a dark defect candidate area are extracted. The feature amount calculation unit 62 calculates the occurrence position, dimensional feature amount (feature feature amount), and luminance feature amount (feature feature amount) for each extracted defect candidate region.

  The defect classification unit 63 determines the type of defect with reference to the defect type classification logic 52 based on the calculated feature amount. The defect type classification logic 52 is manually or automatically created in advance and stored in the data storage unit 5 using a defect sample or the like so as to match the determination result of the inspector using the feature value. ing.

The weighted number calculation unit 64 for each small area divides the entire area of the traveling material into small areas, and calculates the number of bright defects and dark defects for each small area. At this time, not the simple number but the weighted number weighted by the weighting coefficient determined for each defect type is calculated. The weighting coefficient for each defect type is set in advance and stored in the weighting coefficient data storage unit 53. The quality grade determination unit 65 for each small area determines the surface quality for each small area based on the values of the weighted bright defect number and the weighted dark defect number for each small area. The quality grade determination unit 66 for the entire steel plate integrates the surface quality results for each small region, determines the surface quality of the entire traveling material region, and outputs the result to the result output unit 7. The determination of the surface quality is performed with reference to the grade determination logic storage unit 54 that is preset and stored in the data storage unit.
FIG. 2 is a diagram illustrating an example of a processing procedure of a surface quality determination method using the traveling material surface quality determination apparatus according to the present invention.

  First, in an image obtained by imaging the surface of a moving steel sheet, a brighter portion (bright defect candidate region) and a darker portion (dark defect candidate region) than the formation portion (healthy portion) are respectively set to a preset bright portion threshold value and dark portion. Extract by partial threshold (Step 1). The reason why the bright defect candidate area and the dark defect candidate area are separately extracted is to divide the light defect into the light defect and the dark defect and process them separately, as will be described later. There may be one bright part threshold value and one dark part threshold value, but when the extracted defect is classified finely according to the brightness level, two or more of these threshold values are provided. Also good. In the present invention, not only a heavy defect but also a light defect having a low luminance contrast is generally evaluated. Therefore, each threshold value is set lower than that of a conventional surface inspection method.

  Next, the extracted bright defect candidate area and dark defect candidate area are classified into one or more bright defects and one or more dark defects based on the image feature amount (Step 2). Since the harmfulness of defects is closely related to the size and brightness of the defect, the image features are at least dimensional features (defect length, width, area, etc.) and brightness features (average brightness of the defect portion). In addition to this, it is preferable to appropriately include a feature quantity related to the shape and occurrence position of the defect according to the characteristics of the target defect.

  Light defects generally have a very small defect size and tend to have a low luminance contrast. Therefore, it is difficult to accurately classify the type based on the image feature amount. However, light defects caused by the same occurrence cause the same polarity of defects (types of bright and dark defects), although there are variations in the feature quantities such as defect size and brightness. The present invention pays attention to this point, and classifies light defects as light defects and dark defects and processes them separately, so that light defects having different causes of occurrence are not mixed and evaluated. For each classified defect, a weighting coefficient is defined in advance according to the degree of harm.

  For example, bright defect A (weighting factor = 10), bright defect B (weighting factor = 7), bright defect C (weighting factor = 7), bright defect D (weighting factor = 5), bright defect E (weighting factor = 4) ), Bright defect F (weight coefficient = 3), and bright defect G (weight coefficient = 1) (the same applies to dark defects).

  Next, the steel plate surface is divided into small areas of a predetermined dimension of width W × length L, and the weighted addition of the number of bright defects and the number of dark defects in each small area (multiplying the number of each defect type by a weighting coefficient, This is a value obtained by adding all types of defects, and in this specification, the number of weights is individually described (Step 3), where the width W and length L of the small region are the spatial dimensions of light defects in the inspection material. The distribution is set to an appropriate value according to the minimum value of the width and length of the occurrence distribution when light defects are densely formed. The number of light defects in a small area is reduced, and there is a concern about erroneous detection, and conversely, if the values of width W and length L are too large, this may be overlooked when the area of light defect occurrence distribution is small. There is.

  For example, in the case of a steel plate with a plate width of 1000 mm and a plate length of 2000 m, and the expected width and length of the light defect occurrence part are 100 to 1000 mm (full width) and 10 to 2000 m (full length), respectively, the values of W and L Respectively, about 100 mm and about 10 m (about the expected minimum value of the occurrence of light defects) are appropriate. In this way, if the minimum value is set, there is no problem because it can be detected in two or more small areas even if it is distributed over an area of the minimum value or more. Among minor defects, there are ones with a high degree of harmfulness and ones with a small degree of harm, and if these are simply added together in each small area, an appropriate determination result cannot be obtained. In the present invention, as described above, the light defect and the dark defect are each classified finely according to the degree of harmfulness, and these are multiplied by a weighting factor according to the degree of harmfulness, so that the reliability is high. Quality judgment is possible.

  Next, the quality grade of each small area is determined based on the weighted number of bright defects and dark defects (Step 4). For example, when grading is performed in five stages, grade 5 is assigned when the weighted number of light and dark defects is large, grade 1 is assigned when both are small, and grades 4 and 2 are established in the middle. .

  Finally, the surface quality of the steel sheet is determined based on the quality grade of each small area (Step 5). That is, based on the distribution of the quality grade of each small area determined in Step 4 on the steel sheet surface, the nonconforming portion on the steel sheet surface and its degree are determined. Moreover, it is also possible to perform the pass / fail determination for the entire steel sheet based on the determination result.

  In the above description, the case where a single image of the steel sheet surface is processed has been described. However, the same part of the steel sheet surface is captured by two or more types of cameras having different light receiving angles, and surface images obtained from a plurality of cameras are captured. You may make it process. In this case, a bright defect and a dark defect are obtained for each camera image, which is advantageous for determining the type of defect.

  In the above description, the case of inspecting the surface of a steel plate has been described. However, the object of the present invention is not limited to a steel plate, and is applicable to surface inspection of non-ferrous metals such as aluminum plates, paper, and films. It is possible.

  FIG. 3 is a diagram showing an example of a processing procedure in one embodiment of the present invention. The present embodiment is applied to a production line for a hot dip galvanized steel sheet, and the width of the target steel sheet is 800 to 1500 mm and the length is 800 to 3000 m.

  The moving steel plate is imaged with a camera, and defects are extracted from the obtained image by one threshold value for each of the light and dark (Step 11), and a large number including defect area, defect peak luminance, and defect average luminance Are classified into 18 types of light defects and 18 types of dark defects using the defect type classification logic (Step 12). For each classified defect, a weighting factor of 1 to 12 is set in advance as shown in FIG. Divide the steel plate surface into small areas of 200mm width x 20m length, and multiply the number of 8 types of bright defects (bright defect K to bright defect R) in each small area by the weighting factor to obtain the weight number of bright defects. calculate.

  Similarly, the number of dark defects is calculated by multiplying the number of eight types of dark defects (dark defect k to dark defect r) by a weighting factor (Step 13). In this embodiment, a defect having a severity of 6 or more is regarded as a severe defect, and a defect of 5 or less is regarded as a minor defect. If even one severe defect occurs, it is harmful, and only minor defects are treated as harmful. . This is the reason why only defects with a severity of 5 or less are targeted for the evaluation of the defect occurrence density. The grades of the small regions are classified into 6 grades from grade 0 to grade 5 according to the range of the obtained weighted number S1 of light and light defects and weighted number S2 of dark and light defects (Step 14).

  A specific example of the grade determination logic is shown in FIG. Here, P1 to P5 and Q1 to Q5 are generation density evaluation threshold values of bright defects and dark defects, respectively. Grade 0 is a harmless level, grades 1 to 3 are harmful depending on the use of the steel plate, and grades 4 to 5 are harmful levels regardless of the use of the steel plate. For each small area of the steel plate, one with one or more serious defects or one with a minor defect grade of 4 or more (1 or more depending on the application) is removed as a harmful step (Step 15). This method makes it possible to detect light defects and fine oysters due to a defect in the transport roll at an early stage, which is very effective for quality control.

  In Step 15 described above, for example, when there are 3 or more serious defects in the entire steel plate, or when there are 10 or more small regions with a light defect grade of 3 or more, the steel plate may be determined as defective. .

Claims (4)

Image input means for inputting an image of the traveling material surface;
A light / dark region extracting means for extracting a light defect candidate region and a dark defect candidate region from the input image;
Feature quantity calculating means for calculating the dimensional feature quantity and luminance feature quantity of the extracted bright defect candidate area and dark defect candidate area;
Defect classification means for classifying each of the missing candidate regions into one or more bright defects and one or more dark defects using the feature amount;
The generation density of generation density and the dark defect of the bright defect is calculated respectively, and having a determining quality determination unit surface quality from generation density material generation density and dark defects該明defect Traveling material surface quality judgment device. In the running material surface quality determination device according to claim 1,
The quality determination means includes
Define the severity for each type of light and dark defects in advance with a weighting factor,
Dividing the material surface into small areas of a predetermined width and length to obtain a plurality of rectangular small areas ,
For each small area, determine the number of bright defects and the number of dark defects using the weighting factors,
Based on the weighted number of defects weighted number of defects and the dark defect of the bright defect,
A traveling material surface quality judgment device characterized by judging a defect grade for each divided small area.   In the quality determination apparatus for a traveling material according to claim 1 or 2, for a serious defect having a severity of a predetermined value or more, quality is determined for each individual defect separately from the determination based on the defect occurrence density. A quality determination device for a traveling material, characterized in that determination is made. Extracting bright defect candidate areas and dark defect candidate areas from a captured image of the surface of the traveling material according to a preset threshold value for each of the bright part and the dark part,
Classifying the extracted bright defect candidate area and dark defect candidate area into one or more bright defects and one or more dark defects using a feature quantity including a dimension feature quantity and a brightness feature quantity;
The surface quality determination of the running material, characterized by calculating the generation density of the bright defects and the generation density of the dark defects, respectively, and determining the surface quality of the material from the generation density of the bright defects and the generation density of the dark defects Method.