Steel surface inspection method and surface inspection device

Abstract

A surface inspection method for a metal material according to the present invention is a surface inspection method for a metal material of optically detecting a surface defect of the metal material, the surface inspection method including: an irradiation step for irradiating a surface of a metal material with light; an image-capturing step for capturing images of reflected light from the surface of the metal material by the light irradiated in the irradiation step in two or more different wavelength bands to acquire a plurality of images; and a detection step for detecting a surface defect existing on the surface of the metal material from information of relative signal strength between the plurality of images obtained at the same position on the surface of the metal material in the image-capturing step.

Description

[Technical Field]

[0001] The present invention relates to a surface inspection method for a metallic material for optically detecting surface defects of the metallic material, a surface inspection device for a metallic material, and the metallic material. [Background technology]

[0002] In recent years, in the manufacturing process of metal materials, particularly steel products, there has been a demand for detecting surface defects in hot or cold steel products in order to improve yields by preventing mass nonconformity. The term "steel products" as used herein refers to steel products, including steel plates and shaped steel, such as seamless steel pipes, welded steel pipes, hot-rolled steel sheets, cold-rolled steel sheets, and thick plates, as well as semi-finished products, such as slabs, produced during the manufacturing process of these steel products. For this reason, a method for detecting surface defects in steel products has been proposed, which involves irradiating a billet in the seamless steel pipe manufacturing process with light, receiving reflected light, and determining the presence or absence of surface defects based on the amount of reflected light (see Patent Document 1). Another proposed method involves irradiating the hot steel surface with visible light in multiple wavelength ranges, which do not interact with the spontaneous emission of light emitted from the hot steel, from oblique directions symmetrical with respect to the normal to the hot steel surface, obtaining an image of the composite reflected light and an image of the individual reflected light in the normal direction to the hot steel surface, and then detecting surface defects in the hot steel from a combination of these images (see Patent Document 2). In addition, a method has been proposed for detecting surface defects on steel material by irradiating the steel material surface with distinguishable illumination light from directions inclined symmetrically to each other with respect to the normal to the surface, and extracting an arrangement of light and dark areas corresponding to concave shapes from the difference image of two images taken of the inspection target area illuminated from each direction (see Patent Document 3). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 11-37949 [Patent Document 2] Japanese Patent Application Publication No. 59-52735 [Patent Document 3] Patent No. 6079948 Summary of the Invention [Problem to be solved by the invention]

[0004] According to the method described in Patent Document 1, the reflectivity of scale and harmless patterns differs from that of the base steel, so there is a possibility that scale and harmless patterns occurring on sound parts that are not surface defects may be erroneously detected as surface defects. For this reason, the method described in Patent Document 1 distinguishes between surface defects and scale by utilizing the fact that the shape of surface defects (surface flaws) on billets is linear. However, surface defects on steel materials are not limited to linear shapes and can have various shapes, such as circular shapes. For this reason, it is difficult to apply the method described in Patent Document 1 to the detection of surface defects on steel materials. On the other hand, with the method described in Patent Document 2, because there are a huge number of types of surface defects, scale, harmless patterns, etc., it is difficult to distinguish between scale and harmless patterns and surface defects simply by combining images. Furthermore, it is practically difficult to develop detection logic that can handle a huge number of image combinations. Furthermore, the method described in Patent Document 3 solves the above problem and accurately detects concave defects on the surface of steel materials. However, surface defects on steel materials are not limited to concave ones. Defects that do not appear as concaves or irregularities can also occur when rolling is performed with foreign matter attached or when scale forms at concave defect sites. For this reason, it is difficult to detect all surface defects occurring on steel materials using only the method described in Patent Document 3.

[0005] The present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide a surface inspection method and a surface inspection device for metal materials that can detect surface defects of metal materials with high accuracy without omission. Another object of the present invention is to provide a high-quality metal material that is free of surface defects. [Means for solving the problem]

[0006] The surface inspection method for metallic materials according to the present invention is a surface inspection method for metallic materials that optically detects surface defects of a metallic material, and includes an irradiation step of irradiating the surface of the metallic material with light, an imaging step of obtaining multiple images by imaging the light reflected from the surface of the metallic material by the light irradiated in the irradiation step in two or more different wavelength bands, and a detection step of detecting surface defects present on the surface of the metallic material from information on the relative signal intensities between the multiple images obtained from the same position on the surface of the metallic material in the imaging step.

[0007] The detection step may include a step of detecting the surface defects using a classifier created by a machine learning technique using relative intensities between the plurality of images or a plurality of quantities calculated from the relative intensities as feature quantities.

[0008] The irradiating step may include irradiating the light so that the angle with respect to the normal direction of the surface of the metal material is within a range of 60° or more and less than 90°, and the imaging step may include receiving the reflected light so that the light receiving angle with respect to the surface of the metal material is within a range of 0° or more and less than 20°.

[0009] At least one of the two or more different wavelength bands is preferably a wavelength band of 500 nm or less.

[0010] At least one of the two or more different wavelength bands is preferably a wavelength band of 650 nm or more.

[0011] The surface inspection device for metallic materials according to the present invention is a surface inspection device for metallic materials that optically detects surface defects of metallic materials, and comprises an irradiation means that irradiates the surface of the metallic material with light, an imaging means that obtains multiple images by capturing the light reflected from the surface of the metallic material by the light irradiated by the irradiation means in two or more different wavelength bands, and a detection means that detects surface defects present on the surface of the metallic material from information on the relative signal intensities between the multiple images obtained from the same position on the surface of the metallic material by the imaging means.

[0012] The surface properties of the metal material according to the present invention are guaranteed using the surface inspection method for metal material according to the present invention. [Effects of the Invention]

[0013] The surface inspection method and surface inspection device for metal materials according to the present invention can accurately detect all surface defects of metal materials without omission, and can provide high-quality metal materials free from surface defects. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram showing an example of a pattern defect and a harmless pattern. [Figure 2] FIG. 2 is a schematic diagram showing the configuration of the device used in the test. [Figure 3] FIG. 3 is a diagram showing the results of comparing the relationship between signal intensity and wavelength for a pattern defect portion and a healthy portion. [Figure 4] FIG. 4 is a diagram showing the results of comparing the relationship between the incident angle of illumination light, the signal intensity difference between a pattern-like defective portion and a healthy portion, and the wavelength. [Figure 5] FIG. 5 is a schematic diagram showing the configuration of a surface inspection device for metal materials according to the first embodiment of the present invention. [Figure 6] FIG. 6 is a schematic diagram showing the configuration of a surface inspection device for metal materials according to a second embodiment of the present invention. [Figure 7] FIG. 7 is a flowchart showing the flow of a surface inspection process according to one embodiment of the present invention. [Figure 8] FIG. 8 is a diagram showing an example of a difference image. [Figure 9] FIG. 9 is a diagram showing an example of a difference image. DETAILED DESCRIPTION OF THE INVENTION

[0015] The surface of thick steel plates, a type of steel product, is covered with an oxide film called black scale, which can produce harmless patterns that do not affect the quality of the steel product. Furthermore, pattern defects such as scabs also occur on the surface of thick steel plates. However, as shown in Figure 1, it is difficult to distinguish between harmless patterns and pattern defects based on brightness differences. Therefore, the inventors of the present invention focused on the fact that minute red scale (Fe2O3) tends to form on pattern defects, which gives the pattern defects a reddish appearance. They conducted a test to detect pattern defects using spectral reflectance characteristics. Figure 2 shows the configuration of the test equipment. As shown in Figure 2, in this test, a thick steel plate sample SA with a pattern defect was placed on a linear stage 1. A wide-band illumination light L from a xenon light source 2 was irradiated onto the surface of the thick steel plate sample SA, and spectral images at each wavelength were captured using a spectroscopic camera 3 with a one-dimensional field of view.

[0016] Figure 3 shows the results of comparing the relationship between signal intensity and wavelength for pattern defects and healthy areas using spectral images. As shown in Figure 3, the signal intensity characteristics on the short and long wavelength sides are relatively different between pattern defects and healthy areas. This suggests that pattern defects, which are difficult to detect using a single wavelength due to the signal of harmless patterns, may be detectable by comparing signal intensities between wavelengths. Therefore, we performed brightness correction on the spectral image in the 415 nm wavelength band (415 nm image) and the spectral image in the 750 nm wavelength band (750 nm image) so that the average brightness value was constant. Figure 8 shows an image obtained by taking the difference in brightness values ​​between the two spectral images. As shown in Figure 8, the signal of the harmless pattern is canceled out by the difference, and the signal of the pattern defect is emphasized.

[0017] From the above, it was confirmed that pattern defects with different spectral reflectance characteristics from healthy areas can be accurately detected by comparing luminance information between multiple different wavelengths. Figure 4 also shows the results of comparing the incidence angle of illumination light L (angle relative to the surface normal vector of the thick steel plate) with the relationship between the signal intensity difference between pattern defects and healthy areas and wavelength. The values ​​20°, 30°, etc. in the legend of Figure 4 refer to the incidence angle. As shown in Figure 4, the larger the incidence angle of illumination light L, the greater the signal intensity difference between the short and long wavelengths. Therefore, in order to efficiently detect the spectral reflectance characteristics of the target, i.e., the difference in color, it is recommended to increase the incidence angle of illumination light L. While the above explanation focuses on pattern defects in thick steel plates, the present invention can also be applied to the detection of surface defects that occur on the surfaces of other metal materials with spectral reflectance characteristics that differ from healthy areas.

[0018] Hereinafter, with reference to FIGS. 5 to 8, a surface inspection device for metal materials, which is one embodiment of the present invention that has been conceived from the above-mentioned technical idea, will be described.

[0019] Fig. 5 is a schematic diagram showing the configuration of a surface inspection device for metallic materials according to a first embodiment of the present invention. As shown in Fig. 5, the surface inspection device for metallic materials 10 according to the first embodiment of the present invention is an apparatus for detecting surface defects in a plate-shaped steel material S transported in the direction of the arrow in the figure. The surface inspection device for metallic materials 10 according to the first embodiment of the present invention comprises, as main components, a light source 11, an encoder and pulse generator 12, an area sensor 13 capable of capturing spectral images in multiple wavelength bands, an image processing device 14, and a monitor 15.

[0020] The light source 11 irradiates the inspection target portion on the surface of the steel material S with illumination light L in accordance with a trigger signal output by the pulse generator each time a certain number of pulse signals are transmitted from the encoder. The light source 11 is preferably positioned so that the irradiation direction of the illumination light L is tilted within a range of 60° to 90° with respect to the normal direction of the surface of the steel material S. This allows for accurate detection of surface defects from the differential image. Although one light source 11 is used in this embodiment, multiple light sources 11 may be used. While a xenon light source is used as the light source 11 in this embodiment, a combination of light sources with broadband characteristics, such as metal halide light sources, halogen light sources, mercury lamps, and incandescent lamps, or light sources with specific narrowband characteristics, such as LEDs and lasers, may be used as long as they contain components in multiple different wavelength bands. Furthermore, in cases where the conveying speed of the steel material S is fast or the position of the steel material S varies significantly from the designated conveying position (pass line), a flash light source or pulsed lighting may be used to prevent blurring in the image.

[0021] The area sensor 13 captures spectral images of the steel material S at approximately the same position in multiple different wavelength bands. The captured spectral images are preferably coaxial, but can also be aligned by image processing. Examples of area sensors include a Bayer type, in which filters that transmit different wavelength bands are attached to each element so that they are nested, and then multiple images are generated. Other examples of area sensors include a prism type, in which a prism and multiple elements are used and adjusted to be coaxial. While an RGB color camera is inexpensive and preferable, a multiband area sensor with two or four or more channels can also be used. Furthermore, depending on the spectral reflectance characteristics of the healthy and surface defects to be distinguished, wavelength-selective filters can be installed in the optical path, such as in front of the area sensor 13 or light source 11, to enhance color contrast and improve detection capabilities. Furthermore, if the difference in spectral reflectance characteristics between healthy and surface defects appears in a narrow band, the wavelength band received by a wavelength-selective filter or the like can be narrowed if there is sufficient light available. Furthermore, when detecting pattern defects, it is preferable that at least one of the different wavelength bands includes a narrow wavelength band of 650 nm or more, and at least one includes a narrow wavelength band of 500 nm or less.

[0022] The area sensor 13 captures spectral images in synchronization with the light source 11 in accordance with a trigger signal output from the pulse generator. The brightness values ​​of each channel of the spectral images are not saturated, except for the spectral images capturing the end of the steel material S. In addition, although a two-channel area sensor capturing light in the long wavelength band and short wavelength band is used in this embodiment, an apparatus configuration in which three or more wavelength bands are captured using three or more channel area sensors may also be used. In addition, it is preferable that the area sensor 13 receives reflected light so that the light receiving angle with respect to the normal direction of the surface of the steel material S is within a range of 0° or more and less than 20°.

[0023] The image processing device 14 detects surface defects in the inspection target area by performing a subtraction process (to be described later) between the spectral images input from each channel of the area sensor 13. The image processing device 14 then outputs to the monitor 15 the spectral images input from the area sensor 13, the spectral images after the subtraction process, and information related to the detection results of surface defects.

[0024] Second Embodiment Fig. 6 is a schematic diagram showing the configuration of a surface inspection device for metallic materials according to a second embodiment of the present invention. As shown in Fig. 6, a surface inspection device 20 for metallic materials according to the second embodiment of the present invention detects surface defects in a plate-shaped steel material S being transported in the direction of the arrow in the figure. The difference from the surface inspection device 10 for metallic materials according to the first embodiment is that the light source 11 is a line light source 21 and the area sensor 13 is a line sensor 22. The use of the line sensor 22 has the advantage that the field of view is limited to one line, which provides more stable optical conditions compared to the first embodiment, but it also has the disadvantage that the steel material S is vulnerable to positional fluctuations from the transport position (path line) during transport.

[0025] The surface inspection devices 10 and 20 for metal materials configured as described above perform the surface inspection process described below to distinguish between pattern-like defect areas and sound areas with harmless patterns in the inspection target area. The pattern-like defect described here refers to a defect that does not appear to have any irregularities on the surface, such as a foreign object being pressed against the surface or scale forming on a concave defect. Furthermore, sound areas with harmless patterns refer to areas that have a surface coating or surface texture with different optical properties from the base steel portion, such as black scale, which is several to several tens of micrometers thick, and are therefore areas that can cause noise in the surface inspection process.

[0026] [Surface inspection process] Fig. 7 is a flowchart showing the flow of surface inspection processing according to one embodiment of the present invention. The surface inspection processing shown in Fig. 7 starts when an execution command for the surface inspection processing is input to the image processing device 14, and the surface inspection processing proceeds to step S1.

[0027] In the processing of step S1, the image processing device 14 performs alignment processing if the positions of the spectral images of different wavelength bands are misaligned on a pixel-by-pixel basis. If the spectral images cannot be captured coaxially, alignment between the spectral images of different wavelength bands is necessary. The alignment method varies depending on the type of misalignment of the spectral images of each wavelength band, and it is recommended to perform processing such as parallel translation, linear transformation, or one-to-one pixel correspondence as necessary. This completes the processing of step S1, and the surface inspection processing proceeds to processing of step S2.

[0028] In the process of step S2, the image processing device 14 performs first preprocessing such as correction for making the average brightness value constant for a plurality of spectral images of different wavelength bands, brightness unevenness correction, signal intensity normalization, etc. This completes the process of step S2, and the surface inspection process proceeds to the process of step S3.

[0029] In the process of step S3, the image processing device 14 compares multiple spectral images of different wavelength bands and generates a composite processed image that emphasizes only pattern defects by utilizing the differences in spectral reflectance characteristics. Specifically, the image processing device 14 selects two spectral images with large differences in spectral reflectance characteristics and generates a composite processed image by calculating the difference or ratio in luminance between the two spectral images. A case where a difference image of two spectral images is generated as a composite processed image will be described. In this case, the image processing device 14 calculates the luminance value Id1(x,y) of the differential image Id1 by subtracting the luminance value Ir(x,y) of the second spectral image Ir (e.g., the luminance value of the second channel having sensitivity characteristics corresponding to a wavelength band on the long wavelength side) from the luminance value Ib(x,y) of the first spectral image Ib (e.g., the luminance value of the first channel having sensitivity characteristics corresponding to a wavelength band on the short wavelength side), as shown in the following mathematical formula (1).

[0030]

number

[0031] The spectral images Ib and Ir are images with a pixel count of X × Y. In addition, in an xy coordinate system with two orthogonal axes set for each of the spectral images Ib and Ir, the x coordinate is 1≦x≦X and the y coordinate is 1≦y≦Y. In this example, a difference image is generated as the composite processed image, but an image of the ratio of brightness values ​​using the following formula (2) may also be generated as the composite processed image. Furthermore, a composite processed image may also be generated by performing threshold processing separately on the brightness values ​​of the two spectral images and then performing AND processing.

[0032]

number

[0033] Although this process can be performed using only two spectral images, the following methods (a) to (c) are particularly effective for three or more spectral images. This completes the process in step S3, and the surface inspection process proceeds to step S4.

[0034] (a) Spatial transformation method This is an effective method for three-channel, particularly color images, and generates an image that extracts the influence of spectral reflectance characteristics (color tone) by converting the color space. The color space after conversion can be HSV or HLS, and the spectral reflectance information appears in the hue information of each. Alternatively, it can be converted to XYZ, L*u*v*, or L*a*b* space, and after normalizing the luminance component, threshold processing can be performed on the luminance values ​​of each color image.

[0035] (b) Statistical multivariate analysis This is a method in which the luminance values ​​of each wavelength band for each pixel are used as feature vectors, and statistical multivariate analysis is used to extract areas with different spectral reflectance characteristics. As an example, the target inspection area is divided into sufficiently large sections, and in each section, the luminance values ​​of each wavelength band for all pixels are used as feature vectors to perform principal component analysis (PCA), and the image is reconstructed using the Mahalanobis distance of each pixel as its representative value. Since pattern-like defect areas have different colors, it is expected that the Mahalanobis distance will be larger than that of healthy areas. While principal component analysis has been explained here, similar effects can be obtained by similarly calculating the degree of deviation from the model using a Gaussian mixture model, independent component analysis, or regression model.

[0036] (c) Machine learning methods This method is the same as statistical multivariate analysis up to the point where the brightness values ​​of each wavelength band for each pixel are used as a feature vector, but it is a method in which each feature vector is pre-assigned a teacher characterization of whether it is a pattern defect or a healthy area, and a judger is created using general supervised learning to judge for each pixel whether it is a pattern defect or a healthy area. Methods for clustering features may also use k-means, kernel methods, decision tree methods, Gaussian mixture models, regression models, etc. When using this method, threshold processing, which will be described later, is not required, and pixels that are defect candidates can be directly extracted.

[0037] In the process of step S4, the image processing device 14 performs second preprocessing using a frequency filter or the like on the composite processed image to generate an image in which surface defects are emphasized. This completes the process of step S4, and the surface inspection process proceeds to the process of step S5.

[0038] In the process of step S5, the image processing device 14 generates a binarized image by performing threshold processing on the luminance values ​​of the image obtained by the process of step S4. This completes the process of step S5, and the surface inspection process proceeds to the process of step S6.

[0039] In the process of step S6, the image processing device 14 performs processes such as expansion and contraction on the binarized image as necessary to remove connected and isolated points, and then performs a labeling process in which adjacent pixels are regarded as blobs (lumps) and labeled. The image processing device 14 then regards the blobs extracted by the labeling process as surface defect candidates. This completes the process of step S6, and the surface inspection process proceeds to the process of step S7.

[0040] In step S7, the image processing device 14 determines the type and grade (including harmlessness) of each surface defect candidate obtained in step S6. The determination method may involve manually defining the determination rules, or it may be automatically generated using a common machine learning method that uses feature quantities, such as a regression model (linear regression, logistic regression, multiple regression, support vector machine, nonlinear kernel, etc.), a decision tree model, a random forest, a Bayesian estimation model, a Gaussian mixture model, or a boosted version of these models. In this case, by adding representative values ​​(average, maximum) of the brightness values ​​of multiple spectral images and values ​​that allow comparisons between the representative values ​​(sum, difference, ratio), etc., to the feature quantities, determination that takes into account spectral reflectance characteristics is possible. Furthermore, if a sufficient number of data points is obtained, a convolutional neural network may be used. Multiple machine learning methods may also be combined. The determination results are displayed on the monitor 15 for guidance, or are collected in a server and used to determine whether the steel product S can be shipped or whether maintenance is required. This completes step S7, and the surface inspection process ends.

[0041] As is clear from the above description, in the surface inspection process according to one embodiment of the present invention, light is irradiated onto the surface of the steel material S, and the light reflected from the surface of the steel material S by the irradiated light is captured in two or more different wavelength bands. Surface defects present on the surface of the steel material S are detected based on information on the relative signal intensities between multiple images obtained from the same position on the surface of the steel material S. This allows for accurate detection of all surface defects in the steel material S. Furthermore, the surface inspection process according to one embodiment of the present invention is used to investigate the presence or absence of surface defects in metal materials such as the steel material S, and quality assurance is performed to confirm whether the surface defect occurrence status (occurrence rate, defect size, etc.) is below a predetermined tolerance level. This makes it possible to provide metal materials with guaranteed surface properties. In practice, quality can be guaranteed, for example, by stating on a delivery note, inspection certificate, etc. whether the surface defect occurrence status is below a predetermined tolerance level. [Example]

[0042] Example 1 Example 1 shows an example of detecting surface defects on a thick steel plate using the present invention. The surface defects are pressure-bonded to the steel plate surface during the steel plate manufacturing process, and no irregularities are present when viewed from the steel plate surface. Spectral images of the surface defects on the thick steel plate were captured in the 415 nm wavelength band and the 750 nm wavelength band, and brightness correction was performed to maintain a constant average value. Figure 8 shows the difference image. Signals of harmless patterns such as black scale are canceled by the difference, and only the signals of pattern defects are emphasized. By taking advantage of the spectral reflectance characteristics of the surface defect area and the healthy area, it was confirmed that surface defects can be accurately detected by generating a difference image between a spectral image captured with light on the long wavelength side and a spectral image captured with light on the short wavelength side. Depending on the manufacturing conditions of the thick steel plate, surface scale with a surface reflectance spectrum similar to that of the surface defect area may be generated on the healthy area. However, by using a classifier created by machine learning to discriminate the surface defect area extracted from the difference image, overdetection can be suppressed and surface defects can be accurately detected.

[0043] Example 2 Example 2 shows an example of detecting red scale on a thick steel plate using the present invention. The surface of the thick steel plate is unevenly covered with red scale, black scale, and areas of base steel where the scale has peeled off. The scale on hot-worked steel becomes layered as oxidation progresses and is often composed of wüstite (FeO), magnetite (Fe3O4), and hematite (Fe2O3), in that order, from closest to the base steel. When the surface scale is primarily composed of wüstite or magnetite, it tends to become black scale, which is uniform and has high mechanical strength and is resistant to peeling. On the other hand, when the surface scale is primarily composed of hematite, it tends to become red scale, which is non-uniform and has low mechanical strength and is easily peeled off. Red scale is often avoided because it causes problems with workability, and its tendency to peel off can cause stains on factory equipment and lead to uneven color tones in the final product. Therefore, it is sometimes treated as a type of surface defect. Surface images of thick steel plates captured with an RGB color camera were subjected to brightness correction so that the average brightness of the R and B component images was constant. The difference images between the R and B component images are shown in Figures 9(a) and 9(b). The camera element and filter were designed so that the R channel could capture images in a narrow wavelength band of 650 nm or more, and the B channel could capture images in a narrow wavelength band of 500 nm or less. As shown in Figures 9(a) and 9(b), the signals from the black scale and steel substrate were canceled by the difference, and only the red scale signal was emphasized. Based on the above, this invention confirmed that red scale can be accurately detected by taking advantage of the characteristics of the spectral reflectance spectra of the red and black scales to create a difference image between a two-dimensional image captured with long-wavelength light and a two-dimensional image captured with short-wavelength light. In actual operation, quality assurance can be performed by calculating the proportion and extent of red scale occurrence on the entire surface of thick steel plates, and the distribution of red scale occurrence can be used for temperature control in the rolling process and abnormality detection in descaling equipment.

[0044] Although the present invention has been described above as an embodiment, the present invention is not limited to the descriptions and drawings that form part of the disclosure of the present invention. In other words, other embodiments, examples, and operational techniques that can be made by those skilled in the art based on the present invention are all included in the scope of the present invention. [Industrial Applicability]

[0045] According to the present invention, it is possible to provide a surface inspection method and a surface inspection device for metal materials that can detect surface defects of metal materials with high accuracy without omission, and also to provide high-quality metal materials that are free from surface defects. [Explanation of symbols]

[0046] 1 linear stage 2 xenon light source 3 Spectroscopic camera 10,20 Metallic material surface inspection equipment 11 Light source 12 Encoder, pulse generator 13 Area Sensor 14 Image processing device 15 monitors 21 Line light source 22 Line sensor L illumination light S steel material SA thick steel plate sample

Claims

1. A surface inspection method for steel material for optically detecting pattern defects on the surface of steel material, comprising: an irradiation step of irradiating a surface of the steel material with one light including components of a plurality of wavelength bands from one irradiation means; an imaging step of obtaining a plurality of images by imaging, with one imaging means, light reflected from the surface of the steel material by the light irradiated in the irradiation step in two or more different wavelength bands; a detection step of generating a difference image of two spectroscopic images captured in different wavelength bands from among a plurality of images obtained from the same position on the surface of the steel material in the imaging step, and detecting the pattern defect present on the surface of the steel material from the difference image; Including, One of the two spectral images is captured in a wavelength band of 650 nm or more, and the other is captured in a wavelength band of 500 nm or less. Steel surface inspection method.

2. 2. The steel surface inspection method according to claim 1, wherein the detection step includes a step of detecting the pattern defect from the difference image using a classifier created by a machine learning technique using relative intensities between the plurality of images or a plurality of quantities calculated from the relative intensities as feature quantities.

3. A surface inspection device for steel material that optically detects pattern defects on the surface of steel material, One irradiation means for irradiating the surface of the steel material with one light containing components of a plurality of wavelength bands; one imaging means for capturing images of light reflected from the surface of the steel material by the light irradiated by the irradiating means in two or more different wavelength bands, thereby obtaining a plurality of images; a detection means for generating a difference image of two spectroscopic images captured in different wavelength bands from among a plurality of images obtained from the same position on the surface of the steel material by the imaging means, and detecting the pattern defect present on the surface of the steel material from the difference image; Equipped with One of the two spectral images is captured in a wavelength band of 650 nm or more, and the other is captured in a wavelength band of 500 nm or less. Steel surface inspection device.

Images