Surface defect detection method and surface defect detection device

Abstract

The surface defect detection method for optically detecting a surface defect of a belt-shaped body comprises: an image acquisition step for detecting reflected light from the belt-shaped body obtained by illuminating the surface of the belt-shaped body, and acquiring a plurality of images including the surface of the belt-shaped body by relatively scanning and photographing the surface of the belt-shaped body; an average image calculation step of calculating an average image of the plurality of acquired images; an image correction step of performing shadow correction on each of the plurality of acquired images by using the average image to obtain a corrected image; and a defect detection step of detecting a surface defect of the belt-shaped body on the basis of the corrected image, the average image calculation step including identifying, in each of the plurality of images, an inspection target region in which the belt-shaped body is present in the image and causing only pixels in the inspection target region to contribute to the average image.

Description

Technical Field

[0001] The present disclosure relates to a surface defect detection method and a surface defect detection device for detecting surface defects of a strip-shaped object including a steel plate or the like. Background Art

[0002] In recent years, in the manufacturing process of steel products, from the viewpoint of suppressing a large number of defects and improving the yield, it is required to detect the surface defects of steel materials by heat or cold. The steel materials described in the present disclosure include steel products represented by steel plates and steel sections such as seamless steel pipes, welded steel pipes, hot-rolled steel plates, cold-rolled steel plates, and thick plates, as well as semi-finished products such as slabs generated in the process of manufacturing these steel products. As a surface inspection method for optically detecting the surface defects of steel materials, for example, the techniques described in Patent Documents 1 to 3 are conventionally known.

[0003] For example, Patent Document 1 discloses a surface defect detection method that utilizes two distinguishable light sources to illuminate the same inspection object portion of steel from different directions, and acquires an image based on the reflected light of each illumination light, and detects surface defects of the inspection object portion by performing differential processing between the acquired images.

[0004] As a general technique for optically detecting surface defects of steel materials, not limited to the prior art described in Patent Document 1, a method of detecting defects by illuminating the surface of the steel material to be inspected, detecting the reflected light with a camera having a two-dimensional or one-dimensional light-receiving element, and processing the obtained image is widely implemented. As a method for detecting defects from an image, a method is generally used in which the brightness of each pixel of the image is compared with a predetermined threshold value to identify whether each pixel is a sound part or a defective part.

[0005] The following two points can be cited as characteristics of the inspection of surface defects of steel materials.

[0006] First, many steel products are long, so the point is to scan the entire steel and repeatedly take pictures in order to inspect the entire steel. In order to achieve such scanning, the lighting and camera are generally fixed to the production line, etc., and the structure of the steel conveyed in the production line is repeatedly photographed.

[0007] Secondly, since the size of the manufactured steel products, such as product width, varies, the point is that the area where the steel products are shown in the image captured by a fixed camera changes. The area in the image where the steel products are not shown does not need to be inspected, and it is necessary to suppress the situation where defects are mistakenly detected from such areas. Therefore, it is necessary to dynamically identify the range where the steel products are shown in the captured image, that is, the inspection target area, and detect defects only from the inspection target area.

[0008] In the original image directly shot with a camera, in many cases, "shadows" with different brightness levels are generated depending on the position in the image due to the influence of uneven lighting, reduced brightness in the peripheral area of ​​the camera's field of view due to lens characteristics, etc. Therefore, it is difficult to accurately distinguish between healthy parts and defective parts in the method of simply comparing the brightness value of the captured image with a threshold. Therefore, the original image is pre-processed to correct the brightness unevenness called shadow correction to make the brightness level of the healthy part uniform.

[0009] More specifically, correction is performed by dividing the captured image by the shadow pattern or subtracting the shadow pattern from the captured image using a waveform or image representing the shadow pattern. The shadow pattern generally includes a fixed pattern and a dynamic pattern.

[0010] In the case of a fixed pattern, the same pattern is always used for shading correction. The pattern is obtained in advance by experiments, for example. For example, there is a method of using an average image obtained by taking multiple images of steel without defects as the shading pattern.

[0011] On the other hand, in the case of a dynamic pattern, the shadow pattern is calculated or updated based on one or more images captured during the inspection, while the shadow correction is performed on the captured image. By calculating or updating the pattern in accordance with the changes in the shadow of the captured image, the uniformity after correction can be improved compared to a fixed pattern.

[0012] As a basic method of the shadow correction method using such a dynamic pattern, there is a method of correcting the original image using a blurred image subjected to a moving average process or a low-pass filter process on a captured one-dimensional or two-dimensional image as a shadow pattern. There is also a method of obtaining a shadow pattern by approximating the captured image using a polynomial or the like. Improved methods of these are disclosed in Patent Documents 2 and 3.

[0013] For example, Patent Document 2 discloses a shadow correction method for a surface inspection device, which generates a threshold signal based on a first shadow correction signal generated from a scanning detection signal stored in advance up to the previous scan, and uses the first shadow correction signal to replace the signal value relative to the position of the defective part contained in the scanning signal during the current scan to obtain a corrected scanning signal in real time, thereby normalizing the scanning detection signal during the current scan.

[0014] For example, patent document 3 discloses a method, namely: detecting the image area of ​​the inspection object by detecting the boundary between the inspection object and the outside of the inspection object based on the image from the camera, applying a low-pass filter to the image from the camera to extract the shadow component as an image of the same size as the image from the camera, and then, for the image area part of the inspection object in the image, approximating the brightness change of each line using a curve, or approximating the brightness change of the entire image area part of the inspection object using a surface, reconstructing it into an image of the same size with the value of the approximate curve or the approximate surface as the brightness, and correcting the shadow by dividing or subtracting the image from the camera by the reconstructed image.

[0015] Patent Document 1: Japanese Patent Application Publication No. 2016-224069

[0016] Patent Document 2: Japanese Patent Application Publication No. 2012-112729

[0017] Patent Document 3: Japanese Patent Application Publication No. 2004-226347

[0018] However, the above-mentioned existing shading correction methods each have problems.

[0019] For example, in a correction method based on a fixed pattern, it is difficult to perform sufficient correction when the shadow changes greatly, and it is difficult to improve the detection accuracy.

[0020] For example, in the method of obtaining the shadow pattern by moving average processing or low-pass filtering of the captured image, there is a problem of "returning" around the defect in the correction image. In large defects, there is also a problem of "missing" in the center of the defect due to the weakening of the signal due to the correction effect. In any case, the size and area of ​​the defect will change, so there is a possibility that the determination accuracy will deteriorate at the stage of determining the type and harmfulness of the defect after detection.

[0021] In the method described in Patent Document 2, the current scanning signal is corrected using a first shading correction signal based on a previous scanning signal. Therefore, if the edge position of the inspection object material projected in the image moves due to the meandering or width change of the inspection object material, the current scanning signal is corrected using the first shading correction signal in the area where the edge position has changed, resulting in a problem that normal shading correction is difficult to perform near the edge.

[0022] In the method described in Patent Document 3, if the approximation polynomial assumed in the process of generating a shadow pattern by approximating a curve or a curved surface of an original image is inappropriate, the shadow of the original image deviates from the shadow pattern, which makes it impossible to make the brightness of the healthy part uniform. Summary of the invention

[0023] The present disclosure is made in view of the above-mentioned problems, and its purpose is to provide a surface defect detection method and a surface defect detection device that can detect surface defects of a strip-shaped body with higher accuracy. For example, its purpose is to provide a surface defect detection method and a surface defect detection device that suppress "folding back" around a defect portion or "missing" of a large defect, and appropriately perform shading correction until near the edge even when the edge position reflected in the image changes due to the meandering or width change of the inspection object material.

[0024] The present invention discloses a surface defect detection method.

[0025] (1) A surface defect detection method for optically detecting surface defects of a strip-shaped object, comprising:

[0026] an image acquisition step of detecting reflected light from the strip-shaped body obtained by illuminating the surface of the strip-shaped body, and relatively scanning and photographing the surface of the strip-shaped body to acquire a plurality of images including the surface of the strip-shaped body;

[0027] an average image calculation step, calculating the average image of the acquired multiple images;

[0028] an image correction step, using the average image to perform shading correction on each of the plurality of images acquired to obtain a corrected image; and

[0029] a defect detection step of detecting surface defects of the strip-shaped body based on the corrected image,

[0030] The average image calculation step includes identifying, in each of the plurality of images, an inspection target region in which the strip-shaped body exists in the image and causing only pixels in the inspection target region to contribute to the average image.

[0031] (2) In the surface defect detection method described in (1) above,

[0032] The image correction step may also include dividing each of the plurality of images acquired by the average image.

[0033] (3) In the surface defect detection method described in (1) above,

[0034] The image correction step may also include subtracting the average image from each of the plurality of images acquired.

[0035] (4) In the surface defect detection method described in any one of (1) to (3) above,

[0036] The average image calculation step may include identifying, in each of the plurality of images, an unstable portion within the inspection target region in the image and causing only pixels of the stable portion to contribute to the average image.

[0037] (5) In the surface defect detection method described in any one of (1) to (4) above,

[0038] The strip-shaped body may also include steel.

[0039] The present invention discloses a surface defect detection device.

[0040] (6) A surface defect detection device for optically detecting surface defects of a strip-shaped object, comprising:

[0041] an illumination unit for illuminating the surface of the strip;

[0042] an imaging unit configured to detect reflected light from the strip-shaped body obtained by illuminating the surface of the strip-shaped body with the illumination unit; and

[0043] an image processing unit that processes a plurality of images including the surface of the strip-shaped body obtained by scanning and photographing the surface of the strip-shaped body relatively with the illumination unit and the photographing unit to detect surface defects of the strip-shaped body,

[0044] The image processing unit calculates an average image of the plurality of images.

[0045] The image processing unit obtains a corrected image obtained by performing shading correction on the plurality of images using the average image.

[0046] The image processing unit detects surface defects of the strip-shaped body based on the corrected image.

[0047] The image processing unit recognizes, in each of the plurality of images, an inspection target region in which the strip-shaped body exists in the image, and causes only pixels in the inspection target region to contribute to the average image.

[0048] According to a surface defect detection method and a surface defect detection device according to an embodiment of the present disclosure, it is possible to detect surface defects of a strip-shaped body with higher accuracy. For example, "folding back" around a defect portion or "missing" of a defect with a large area is suppressed. Furthermore, even if the position of the edge reflected in the image changes due to the meandering or width change of the inspection object material, shading correction can be appropriately performed until near the edge. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram showing the structure of a surface defect detection device involved in one embodiment of the present disclosure.

[0050] Figure 2 It means by Figure 1 Flowchart of image processing performed by the image processing unit.

[0051] Figure 3 Yes means Figure 1 Figure 4 shows multiple images of steel.

[0052] Figure 4 It is a figure which shows the Example of this disclosure.

[0053] Figure 5 It is a diagram showing a comparative example based on the prior art.

[0054] Figure 6 It is a diagram showing a comparative example based on another conventional technology.

[0055] Fig. 7A It shows Figure 4 (d) is a graph of the corrected brightness waveform between A and A in the shading correction image.

[0056] Figure 7B It shows Figure 5 (d) is a graph of the corrected brightness waveform between i-i in the shading correction image.

[0057] Figure 7C It shows Figure 6 (d) is a graph of the corrected brightness waveform between U and U in the shading correction image. DETAILED DESCRIPTION

[0058] Hereinafter, the configuration and operation of a surface defect inspection device according to an embodiment of the present disclosure will be mainly described with reference to the drawings.

[0059] Figure 1 1 is a schematic diagram showing the structure of a surface defect detection device 1 involved in one embodiment of the present disclosure. The surface defect detection device 1 optically detects surface defects of a strip-shaped body. In the present disclosure, "strip-shaped body" includes, for example, steel materials, etc. The steel materials described in the present disclosure include, for example, steel products represented by seamless steel pipes, welded steel pipes, hot-rolled steel plates, cold-rolled steel plates, thick plates and other steel plates and steel sections, as well as semi-finished products such as slabs generated in the process of manufacturing these steel products.

[0060] like Figure 1As shown, a surface defect detection device 1 according to an embodiment of the present disclosure detects surface defects of a steel material PL as a strip conveyed on a conveyor roller table TB in the direction of arrow D in the figure. The surface defect detection device 1 has lighting devices 2a, 2b, area sensors 3a, 3b, a processing device 4 and a display device 5 as main components. The lighting devices 2a, 2b correspond to the "lighting unit" described in the scope of protection requested in this application. The area sensors 3a, 3b correspond to the "shooting unit" described in the scope of protection requested in this application.

[0061] The lighting devices 2a and 2b have any light emitting element for illuminating the surface of the strip. For example, each time the steel material PL moves a certain distance in the direction of arrow D on the conveyor roller table TB, the lighting devices 2a and 2b emit light to illuminate the inspection position on the surface of the steel material PL.

[0062] The area sensors 3a and 3b have an arbitrary light receiving element that detects reflected light from the strip-shaped body obtained by illuminating the surface of the strip-shaped body by the lighting devices 2a and 2b. The area sensors 3a and 3b include, for example, cameras. For example, the area sensors 3a and 3b detect reflected light from the inspection position on the surface of the steel material PL in synchronization with the light emission of the lighting devices 2a and 2b, respectively, and capture an image of the surface of the steel material PL.

[0063] exist Figure 1 In the illustrated embodiment, the surface defect detection device 1 has two sets of lighting devices and area sensors, but is not limited thereto. The surface defect detection device 1 may also have one or more than three sets of lighting devices and area sensor groups, for example, depending on the product width of the steel material PL to be inspected. The surface defect detection device 1 may be configured with multiple lighting devices for one area sensor, or multiple area sensors for one lighting device.

[0064] The operation processing device 4 has one or more processors. In one embodiment, the "processor" is a general-purpose processor or a dedicated processor customized for a specific process, but is not limited thereto. The operation processing device 4 acquires the two-dimensional image captured by the area sensors 3a and 3b to detect the surface defects of the steel material PL. The operation processing device 4 has image processing units 41a and 41b and a comprehensive processing unit 42 inside.

[0065] The image processing units 41a and 41b detect surface defects of the strip by processing a plurality of images including the surface of the strip obtained by scanning and photographing the surface of the strip relatively using the lighting devices 2a and 2b and the area sensors 3a and 3b. The image processing units 41a and 41b detect the existence range of the steel material PL and the surface defects on the steel material PL reflected in the field of view of each area sensor by performing a series of image processing described later. The image processing units 41a and 41b can also determine the type and harmfulness of the defect based on the characteristics of each detected surface defect.

[0066] The comprehensive processing unit 42 collects information on the existence range of the steel material PL and information on surface defects detected by the image processing units 41a and 41b respectively over the entire steel material PL. The comprehensive processing unit 42 creates inspection performance information such as a defect map showing where defects exist on the steel material PL, and a summary table summarizing the number of defects on the steel material PL by defect type and harmfulness. The comprehensive processing unit 42 outputs the created inspection performance information to the display device 5, and displays it to the operator using the display device 5.

[0067] The display device 5 has one or more output interfaces for outputting information to present to the operator. For example, the display device 5 has a display that outputs information in the form of images.

[0068] Next, refer to Figure 2 The image processing performed by the image processing units 41 a and 41 b will be described.

[0069] Figure 2 It means by Figure 1 The flowchart of the image processing performed by the image processing units 41a and 41b is shown in FIG. The steps of the image processing performed by the image processing units 41a and 41b are the same. Therefore, the image processing in the system of the lighting device 2a, the area sensor 3a, and the image processing unit 41a will be described as an example below.

[0070] In step S1, the image processing unit 41a acquires an input image I from the area sensor 3a. n (x, y). Input image I n (x, y) is an image obtained by the area sensor 3a detecting and capturing the reflected light from the surface of the steel material PL illuminated by the lighting device 2a in synchronization with the light emission of the lighting device 2a. Here, the subscript n represents the nth capture since the inspection of the steel material PL started. (x, y) represents the two-dimensional coordinates of the input image. x is the width direction of the steel material PL, i.e., the width direction of the steel material PL. Figure 1 y is the direction perpendicular to the direction of the arrow D in the figure. Figure 1 The direction of the arrow D in the image corresponds to the coordinates. nThe value of (x, y) represents the brightness at each coordinate (x, y).

[0071] In step S2, the image processing unit 41a processes the input image I acquired in step S1. n (x, y) to detect the inspection object area, that is, in the input image I n The range where the steel material PL exists in (x, y) The method for detecting the inspection target area may be any method suitable for the inspection target material and its background.

[0072] For example, in the input image I n When the steel material PL at (x, y) is bright enough relative to the background, a method is adopted in which a threshold value T is used to distinguish between the background and the inspection target material. R At each coordinate (x, y), the binary image R representing the inspection object area is calculated by the following equations (1) and (2): n (x, y).

[0073] in I n (x, y) ≥ T R When R n (x, y) = 1 (inspection target area) (1)

[0074] in I n (x, y)<T R When R n (x, y) = 0 (background) (2)

[0075] In addition, the following method can also be used, namely: in the input image I n Search along the x direction at each y coordinate of (x, y) and n (x, y) The first time that the specified number of pixels exceeds T R The position of is recognized as the boundary between the steel material PL and the background to detect the inspection object area.

[0076] In step S3, the image processing unit 41a converts the input image I detected in step S2 into n The brightness of the pixels included in the inspection target area of ​​(x, y) is added to the cumulative brightness image S(x, y). In the cumulative brightness image S(x, y), the brightness of the pixels from the input image I1(x, y) to I n-1 The image processing unit 41a updates the cumulative brightness image S(x, y) as shown in the following equation based on the brightness values ​​in the inspection target area of ​​each image from (x, y) to (x, y).

[0077] S(x, y):=S(x, y)+I n (x, y) × R n (x, y) (3)

[0078] In step S4, the image processing unit 41a processes the input image I detected in step S2. n The count image C(x, y) is incremented by 1 for the pixels included in the inspection target area at coordinates (x, y). The pixels at coordinates (x, y) are accumulated in the count image C(x, y) from the input image I1(x, y) to I n-1 The image processing unit 41a updates the count image C(x, y) according to the following equation based on the number of times (x, y) falls within the inspection target area.

[0079] C(x,y):=C(x,y)+R n (x, y) (4)

[0080] The image processing unit 41a initializes the count image C(x, y) for each steel material PL as the inspection target material.

[0081] In step S5, the image processing unit 41a determines whether there is an input image that has not been acquired. If the image processing unit 41a determines that there is an input image that has not been acquired, the process returns to step S1 and processes the next input image I. n+1 (x, y) The processing of steps S1 to S4 is executed again. If the image processing unit 41a determines that there is no unacquired input image, the processing of the next step S6 is executed.

[0082] In step S6, the image processing unit 41a calculates the average image A(x, y) of the plurality of input images acquired in step S1. More specifically, the image processing unit 41a divides the cumulative brightness image S(x, y) by the count image C(x, y) to calculate the average image A(x, y). The image processing unit 41a performs the calculation as shown in the following formula at each coordinate (x, y).

[0083] When C(x, y)≠0, A(x, y):=S(x, y) / C(x, y)(5)

[0084] (When C(x, y) = 0, the value of A(x, y) is arbitrary.)

[0085] The value of the average image A(x, y) calculated by formula (5) is the average brightness of the steel material PL at a position corresponding to the coordinates (x, y) of the input image within the field of view of the area sensor 3a, and shows the shadow pattern caused by the unevenness of the lighting device 2a as a whole. Here, in the average image calculation step of step S6, the image processing unit 41a identifies the inspection object area where the strip exists in each of the multiple input images and only allows the pixels in the inspection object area to contribute to the average image A(x, y). It should be noted that in each input image, the brightness of the input image contributes to the average value only when the coordinates (x, y) are within the inspection object area, and the contribution from the background is excluded.

[0086] In step S7, the image processing unit 41a uses the average image A(x, y) to perform shading correction on each of the plurality of images acquired in step S1 and acquires a corrected image. More specifically, the image processing unit 41a uses the average image A(x, y) calculated in step S6 to perform shading correction on the input image I n Shading correction for (x, y).

[0087] In the image correction step of step S7, the image processing unit 41a divides each of the multiple images acquired in step S1 by the average image A(x, y). More specifically, if the input image I n The image (x, y) after shading correction is set as J n (x, y), the image processing unit 41a performs calculation as shown in the following equation in the division type shading correction.

[0088] J n (x, y): =I n (x, y) / A(x, y) (6)

[0089] Alternatively, in the image correction step of step S7, the image processing unit 41a subtracts the average image A(x, y) from each of the plurality of images acquired in step S1. More specifically, the image processing unit 41a performs calculation as shown in the following equation in the subtractive shading correction.

[0090] J n (x, y): =I n (x,y) –A(x,y) (7)

[0091] Among them, in the input image I n (x, y) in the inspection target area, that is, R n The above-mentioned shading correction calculation is performed for the coordinate (x, y) where (x, y) = 1, and the calculation for the other coordinates corresponding to the background (R n(x, y) = 0) is set to an arbitrary value. By this shading correction, the average brightness level of the healthy part of the steel material PL is constant regardless of the position. For example, in the division type shading correction using equation (6), the average brightness level of the healthy part becomes J n (x, y) = 1. For example, in the subtractive shading correction using equation (7), the average brightness level of the healthy part is J n (x, y)=0.

[0092] Here, as described above, in the average image A(x, y), only when the coordinate (x, y) is within the inspection target area in each input image, the brightness of the input image contributes to the average value, and the contribution from the background is excluded. Therefore, even if the steel material PL is bent or the width changes in a series of input images I n In the coordinates (x, y) such as the surface of the steel material PL or the background, shading correction is appropriately performed.

[0093] If you need to adjust the shadow correction image J n If the image processing unit 41a determines the average brightness level of (x, y) or the contrast of the defect signal, the image processing unit 41a may perform appropriate linear conversion or the like. For example, in order to make the image subjected to the division type shading correction based on the formula (6) a grayscale image of 256 gray levels, the image processing unit 41a converts the average brightness level to 128. In this case, the image processing unit 41a uses the converted shading correction image as J n ’ (x, y), and perform the following conversion processing.

[0094] J n ’ (x, y) = J n (x, y)×128 (8)

[0095] In step S8, the image processing unit 41a performs a shading correction on the shading-corrected image J obtained in step S7. n Binarization of (x, y). Binarization is used to determine whether each pixel belongs to a defective part or not. If the binary image is set as B n (x, y), the image processing unit 41a generates the shading-corrected image J n For each pixel of (x, y), the binary image B is calculated as follows by comparing with two thresholds T1 and T2: n (x, y).

[0096] J n (x, y)≤T1 or J n When (x, y) ≥ T2, B n (x, y): = 1 (9)

[0097] T1<J n When (x, y) < T2, B n (x, y): = 0 (10)

[0098] Among them, threshold T1 is a value less than the average brightness level. Threshold T2 is a value greater than the average brightness level. When the shadow correction is a division type, T1<1<T2. When the shadow correction is a subtraction type, T1<0<T2. When a conversion like equation (8) is performed, thresholds T1 and T2 are respectively the values ​​before and after the converted average brightness level. When the defect to be detected is either brighter or darker than the healthy part on the image, the image processing unit 41a may also compare with only one of the thresholds T1 and T2.

[0099] In step S9, the image processing unit 41a uses the binary image B obtained in step S8 n (x, y) and mark the extracted blob. Here, the blob refers to the binary image B n The pixel where the value of (x, y) is 1, that is, the pixel belonging to the defective part is continuous. The marking means giving a label (serial number) for identification to each spot.

[0100] Sometimes in the binary image B n The area of ​​pixels belonging to the same defect in (x, y) is discontinuous and is divided into two or more spots. In response to this situation, the image processing unit 41a can treat the spots as one spot when the distance between the spots is less than a specified distance and assign the same label. Since there are cases where small spots are generated due to harmless patterns on the surface of the steel material PL even in healthy parts, the image processing unit 41a can also remove spots with a small area (number of pixels).

[0101] In step S10, the image processing unit 41a calculates the feature amount of the spot, that is, the defect marked in step S9. n (x, y) (or input image I n (x, y)) and the binary image B n The feature quantity is calculated from the pixel area belonging to each spot (defect) in (x, y). The feature quantity includes, for example, the feature quantity indicating the size of the defect such as width, length, and area, the feature quantity related to the shape of the defect such as aspect ratio and roundness, and the feature quantity related to the density of the defect such as average brightness and brightness histogram.

[0102] In step S11, the image processing unit 41a corrects the image J based on the shading. n(x, y) to detect the surface defects of the strip. More specifically, the image processing unit 41a performs defect determination based on the feature amount of each spot (defect) calculated in step S10. For example, the image processing unit 41a assigns the type and harmfulness of the defect to each spot. As a determination method, a predetermined IF-THEN rule can be applied, a discriminator generated by various machine learning methods can be applied, and a combination of these can be applied.

[0103] In step S12, the image processing unit 41a determines whether there is an input image I that has not been processed from step S7 to step S11. n (x, y) If the image processing unit 41a determines that there is an unfinished input image, the process returns to step S7 and repeats the process from step S7 to step S11. If the image processing unit 41a determines that the process is completed for all input images, the process ends.

[0104] The surface defect detection device 1 performs the processing of the above steps S1 to S12, so that even when the steel material PL meanders and causes the edge position to change, the shadow correction near the plate edge is normally performed, and the shadow correction that does not produce "folding back" around the defective part can be realized. As a result, the surface defect detection device 1 can perform high-sensitivity detection of surface defects and appropriately determine the type and harmfulness of the defects.

[0105] If there is an unstable part such as a defective part in the inspection target area, that is, a part with a higher or lower brightness than the surrounding healthy part, it contributes to the average image A(x, y). For example, at the coordinate (x, y) where the value of the count image C(x, y) is small, its contribution becomes large, and additional distortion is generated in the average image A(x, y) as the estimated shadow pattern. In order to suppress such distortion, in step S2, the image processing unit 41a may calculate R n (x, y) and calculate the next binary image Q n (x, y).

[0106] In R n When (x, y) = 1 and (x, y) is a stable part, Q n (x, y) = 1 (11)

[0107] In R n When (x, y) = 0 or (x, y) is an unstable part, Q n (x, y) = 0 (12)

[0108] In step S3, the image processing unit 41a replaces the equation (3) with the following equation (3') to perform calculation.

[0109] S(x, y):=S(x, y)+In (x, y) × Q n (x,y)(3')

[0110] In step S4, the image processing unit 41a replaces the equation (4) with the following equation (4') to perform calculation.

[0111] C(x,y):=C(x,y)+Q n (x,y)(4')

[0112] In equations (11) and (12), in order to determine whether it is a stable portion or an unstable portion, for example, the image processing unit 41a may use n Alternatively, the image processing unit 41a may calculate the shading-corrected image J by executing steps S1 to S8 once. n (x, y) and binary image B n (x, y) will become B n The pixel (x, y) of (x, y) = 1 is determined as an unstable portion. Then, the image processing unit 41a may return to step S1 again, replace the above equations (3') and (4') in steps S3 and S4, and execute until step S12.

[0113] At this time, in the average image calculation step of step S6, the image processing unit 41a can identify the unstable part in the inspection object area in each of the multiple input images and make only the pixels of the stable part contribute to the average image A(x, y).

[0114] Below, refer to Figure 3 to Figure 7C The following mainly describes the embodiments of the present disclosure and comparative examples based on the prior art.

[0115] Figure 3 Yes means Figure 1 Figure 2 shows multiple images of the steel material PL. Figure 3 The figure shows a part of 61 images obtained by repeatedly photographing at constant intervals in the length direction of a thick steel plate from the edge E to the center of the thick steel plate as the steel material PL being conveyed on the conveyor roller table TB using a camera of the photographing unit fixed above the conveyor roller table TB and flash lighting of the lighting unit. Figure 3 The upper layer (a) shows an image near the front end in the longitudinal direction of the thick steel plate. Figure 3 (b) of the middle layer shows an image of the central portion in the longitudinal direction of the thick steel plate. Figure 3 The lower layer (c) shows an image near the end of the thick steel plate in the longitudinal direction. As shown in the above three images, it can be seen that the position of the edge E of the thick steel plate moves within the camera field of view and meanders on the conveyor roller table TB.

[0116] Figure 4 It is a figure which shows the Example of this disclosure. Figure 4 Express Figure 3 An example of an input image having a stain-like defect and a processed image thereof is shown in FIG. 1 , which is an example of a series of images of a thick steel plate to which the image processing according to the surface defect detection device 1 according to one embodiment of the present disclosure is applied.

[0117] Figure 4 (a) represents the input image I n A graph of (x, y). Figure 4 (b) is a diagram showing the average image A(x, y). Figure 4 The average image A(x, y) of (b) corresponds to the shadow pattern. Figure 4 (c) shows a binary image R representing the inspection object area. n A graph of (x, y). Figure 4 (d) represents the shadow correction image J n A graph of (x, y). Figure 4 (e) represents the shadow correction image J n The binary image B obtained by thresholding (x, y) n A graph of (x, y). Figure 4 The binary image B of (e) n (x, y) is an image where a defective portion is detected.

[0118] exist Figure 4 in Figure 2 In the detection of the inspection object area in step S2, the image processing unit 41a searches for the edge of the thick steel plate from the left side of the input image, and defines the position where more than 20 pixels continuously exceed the specified brightness value to the right as the edge position. Since a halation occurs near the edge, the image processing unit 41a excludes 18 pixels from the edge as an unstable portion from the inspection object area. Figure 2 In the labeling process of step S9, the image processing unit 41a removes spots with an area smaller than 20 pixels as noise.

[0119] Figure 5 It is a diagram showing a comparative example based on the prior art. Figure 5 Expressing Figure 4 The image shown in (a) is the result of image processing when a shading pattern is calculated using a moving average filter and shading correction is performed on the same input image. The shading pattern is calculated for each y coordinate and a moving average filter of 64 pixels is used in the x direction.

[0120] Figure 5 (a) indicates Figure 4 (a) Figure of the same input image. Figure 5 (b) means Figure 5 (a) is a diagram of a shadow pattern generated by applying a moving average filter to the input image. Figure 5 (c) shows the inspection object area Figure 4 (c) Figure 2. The same binary image. Figure 5 (d) is based on Figure 5 (b) A diagram of a shading-corrected image of a shading pattern. Figure 5 (e) means Figure 5 (d) is a diagram of a binarized image obtained by performing threshold processing on the shading-corrected image. Figure 5 The binary image (e) is an image where a defect is detected. Figure 4 Same processing and parameters.

[0121] Figure 6 It is a diagram showing a comparative example based on another conventional technology. Figure 6 Expressing Figure 4 The image shown in (a) is the result of image processing in which a shading pattern is calculated using a second-order polynomial approximation and shading correction is performed on the same input image.

[0122] Figure 6 (a) indicates Figure 4 (a) Figure of the same input image. Figure 6 (b) means according to Figure 6 (a) is a diagram of a shading pattern generated by applying a second-order polynomial approximation to the brightness waveform in the x direction for each y-coordinate of an input image. Figure 6 (c) shows the inspection object area Figure 4 (c) Figure 2. The same binary image. Figure 6 (d) is based on Figure 6 (b) A diagram of a shading-corrected image of a shading pattern. Figure 6 (e) means Figure 6 (d) is a diagram of a binarized image obtained by performing threshold processing on the shading-corrected image. Figure 6 The binary image (e) is an image where a defect is detected. Figure 4 Same processing and parameters.

[0123] 7A to 7C The diagram compares the shading correction results of the embodiment based on the technology of the present disclosure and two comparative examples. Fig. 7A It shows Figure 4(d) is a graph of the corrected brightness waveform between A and A in the shading correction image. Figure 7B It shows Figure 5 (d) is a graph of the corrected brightness waveform between i-i in the shading correction image. Figure 7C It shows Figure 6 (d) is a graph of the corrected brightness waveform between U and U in the shading correction image.

[0124] The embodiment of the present disclosure Figure 4 The average image A(x, y) of (b) shows well Figure 4 (a) Input image I n The shadow of (x, y) is almost unaffected by the patterns and defects on the surface of the thick steel plate. Figure 4 (d) Shadow correction image J n (x, y), such as Fig. 7A As shown in the curve graph, the overall brightness is uniform and no "reflection" occurs around the defective part.

[0125] On the other hand, in the comparative example of the prior art Figure 5 In the shadow correction image of (d), as Figure 7B As shown in the curve graph, slightly brighter areas appear on the left and right of the stain-like defect (dark area), that is, "return" relative to the defect (dark area). Figure 5 In the comparative example shown in (d), although no erroneous detection of the "returning" portion occurs, when the signal of the defective portion is stronger, the signal level of the "returning" portion also increases proportionally, which becomes a cause of erroneous detection.

[0126] As another comparative example of the prior art Figure 6 In the shadow correction image of (d), as Figure 7C As shown in the curve diagram, although no Figure 7B The shading correction is not adequate and is not performed sufficiently. Figure 6 In the binarized image of (e), an erroneous detection occurs at the right end of the image.

[0127] As described above, when the technology of the present disclosure is used, shading correction can be appropriately performed even when meandering or width variation occurs in the inspection target material, and “folding back” around the defective portion can be suppressed, compared to the conventional technology.

[0128] According to the surface defect detection method and surface defect detection device 1 involved in one embodiment as described above, an average image calculated from a plurality of images is used as a shadow pattern, and in each of the plurality of images, only pixels in the inspection object area contribute to the average image, so that the surface defects of the strip-shaped body can be detected with higher accuracy. More specifically, the "return" around the defective part or the "omission" of large-area defects is suppressed in the corrected image. In addition, even if the edge position reflected in the image changes due to the meandering or width change of the inspection object material, the shadow correction can be appropriately performed until near the edge.

[0129] Although the present disclosure is described based on the drawings and embodiments, it should be noted that those skilled in the art may make various modifications and changes based on the present disclosure. Therefore, it should be noted that these modifications and changes are included in the scope of the present disclosure. For example, the functions included in each structure or each step can be reconfigured in a logically non-contradictory manner, and multiple structures or steps can be combined into one or divided.

[0130] For example, the shape, size, arrangement, orientation and number of the above-mentioned structural parts are not limited to the contents of the above-mentioned description and the drawings. The shape, size, arrangement, orientation and number of the above-mentioned structural parts can be arbitrarily configured as long as they can achieve their functions.

[0131] For example, it is also possible to configure a general-purpose electronic device such as a smartphone or a computer to function as the surface defect detection device 1 involved in the above-mentioned embodiment. Specifically, a program describing the processing content of each function of the surface defect detection device 1 involved in the embodiment is stored in the memory of the electronic device, and the program is read and executed by the processor of the electronic device. Therefore, the disclosure involved in one embodiment can also be implemented as a program executable by a processor.

[0132] Alternatively, the disclosure of one embodiment can also be implemented as a non-transitory computer-readable medium, which stores a program executable by one or more processors in order to enable the surface defect detection device 1 and the like involved in the embodiment to perform various functions. It should be understood that they are also included in the scope of the present disclosure.

[0133] In the above-mentioned embodiments and examples, an area sensor, i.e., a camera having a two-dimensional array of imaging elements, etc., is used as the "imaging unit", but the present invention is not limited thereto. A linear sensor, i.e., a camera having a one-dimensional array of imaging elements, etc., may also be used as the "imaging unit". When a linear sensor is used, the imaging of the inspection target material is performed for each line in the X direction. Therefore, the surface defect detection device 1 may also convert the input image I nThe size of (x, y) in the Y direction is set to 1, or the input image I is formed by accumulating a constant number of lines. n (x, y).

[0134] In the above embodiment, the “strip-shaped body” is described as including, for example, steel materials, but the present invention is not limited thereto. The strip-shaped body may include any other strip-shaped or sheet-shaped object other than steel materials.

[0135] Description of Reference Numerals

[0136] 1…surface defect detection device; 2a…illumination device (illumination unit); 2b…illumination device (illumination unit); 3a…area sensor (photographing unit); 3b…area sensor (photographing unit); 4…arithmetic processing device; 41a…image processing unit; 41b…image processing unit; 42…integrated processing unit; 5…display device; E…edge portion; PL…steel material (strip); TB…conveying roller table.

Claims

1. A surface defect detection method for optically detecting surface defects of a strip-shaped object, characterized in that: include: an image acquisition step of detecting reflected light from the strip-shaped body obtained by illuminating the surface of the strip-shaped body, and relatively scanning and photographing the surface of the strip-shaped body to acquire a plurality of images including the surface of the strip-shaped body; an average image calculation step, calculating an average image of the plurality of images acquired; an image correction step, using the average image to perform shadow correction on each of the plurality of images acquired to obtain a corrected image; as well as a defect detection step of detecting surface defects of the strip-shaped body based on the corrected image, The average image calculation step includes identifying, in each of the plurality of images, an inspection target region in which the band-shaped body exists in the image and causing only pixels in the inspection target region to contribute to the average image.

2. The surface defect detection method according to claim 1, characterized in that: The image correction step includes dividing each of the plurality of acquired images by the average image.

3. The surface defect detection method according to claim 1, characterized in that: The image correction step includes subtracting the average image from each of the plurality of acquired images.

4. The surface defect detection method according to any one of claims 1 to 3, characterized in that: The average image calculation step includes identifying, in each of the plurality of images, an unstable portion within the inspection object region in the image and causing only pixels of the stable portion to contribute to the average image.

5. The surface defect detection method according to any one of claims 1 to 3, characterized in that: The strip-shaped body comprises steel material.

6. A surface defect detection device for optically detecting surface defects of a strip-shaped object, characterized in that: have: an illumination unit for illuminating the surface of the strip; a photographing unit configured to detect reflected light from the strip-shaped body obtained by illuminating the surface of the strip-shaped body with the illuminating unit; as well as an image processing unit that detects surface defects of the strip-shaped body by processing a plurality of images including the surface of the strip-shaped body obtained by scanning and photographing the surface of the strip-shaped body relatively with the illumination unit and the photographing unit, The image processing unit calculates an average image of the plurality of images. The image processing unit acquires a corrected image obtained by performing shading correction on the plurality of images using the average image. The image processing unit detects surface defects of the strip-shaped body based on the corrected image. The image processing unit identifies, in each of the plurality of images, an inspection target region in which the strip-shaped body exists in the image, and causes only pixels in the inspection target region to contribute to the average image.

Images