Anomaly detection apparatus and anomaly detection method

Abstract

An abnormality detection device for detecting a rolling abnormality of a surface of a rolled material (1) rolled by a rolling mill, includes a camera (81, 82) that photographs the rolled material (1) as a detection target of an abnormality, and an image processing section (92) that divides luminance data of pixels within a range of the rolled material (1) into three components of R, G, and B values based on an image photographed by the camera (81, 82), and detects the rolling abnormality of the surface based on a relationship between luminance data of two components among the luminance data of the components. Thus, an abnormality detection device and an abnormality detection method that can improve the rolling abnormality detection accuracy of a plate compared to the past are provided.

Description

Technical Field

[0001] The present invention relates to an abnormality detection device and an abnormality detection method. Background Art

[0002] In a production line, as an example of a surface inspection device that accurately determines the texture signal of a defect-free portion and accurately and efficiently sets a judgment threshold, Patent Document 1 describes a surface inspection device that detects defects on the surface of a traveling metal steel strip, wherein the threshold calculation unit includes: a concentration histogram calculation unit that determines the concentration histogram per unit area from an image signal; a standard deviation calculation unit that determines the peak position of the concentration histogram, assumes a normal distribution function with the peak position as the average value, and determines the standard deviation value of the assumed normal distribution function from the density histogram degrees near the peak position; and a threshold setting unit that sets a judgment threshold corresponding to the standard deviation value, wherein the judgment threshold is determined in such a way that the standard deviation value does not contain defect signal components other than texture signal components as much as possible.

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Laid-Open No. 2010-85096 Summary of the Invention

[0006] As for hot rolling finishing mills, with the increase in thin plate / high load rolling, when the tail end of the plate passes after the post-tension is released, the rolled plate extrusion phenomenon (this phenomenon is called rolling abnormality) caused by the rapid winding of the plate becomes more frequent, accompanied by damage to the roll surface, the number of roll replacements increases, and the product yield may decrease.

[0007] While these issues are currently avoided through operator-guided interventions such as bending and leveling, they rely heavily on operator skill. Computer image processing, utilizing the operator's observations of monitored images, allows for rapid intervention as soon as plate extrusion (rolling anomaly) is detected, increasing the probability of preventing subsequent rolling anomalies or equipment failures.

[0008] The rolled sheet squeeze phenomenon is a phenomenon in which a metal strip is rolled in a folded state while passing through a rolling mill.

[0009] If the folded portion of the plate is rolled in this state, the working heat generated in this portion becomes higher than that of the other portions, resulting in a brightness brighter than that of the normal portion of the plate surface, enabling detection of rolling abnormalities accompanied by brightness differences.

[0010] Thus, in hot rolling, in order to detect a rolling abnormality such as a squeeze of a rolled plate, an image of a surface of the plate is monitored using a camera, and when the brightness of the plate surface exceeds a certain threshold and the area becomes large, the plate abnormality can be detected.

[0011] For example, in Patent Literature 1, when the distribution area of white pixels in an image obtained after performing 2-value processing on a tail end image detected using a tail end detection unit is equal to or greater than a prescribed area, it is detected that a squeeze has occurred at the tail end.

[0012] In the technology described in the above Patent Literature 1, the image is judged using the shades of white and black, but according to the research of the present inventor, the brightness distribution of an actual image is very different from white and black, and there is room for detecting abnormalities with high precision.

[0013] The present application aims to provide an abnormality detection device and an abnormality detection method that can improve the precision of detection of a rolling abnormality of a plate compared to the past.

[0014] The present application includes a plurality of means for solving the above problem, but if one example is given, an abnormality detection device that detects a rolling abnormality of a surface of a metal strip plate rolled by a rolling mill is provided, the abnormality detection device being characterized by including: a camera that photographs the metal strip plate that is the detection target of an abnormality; and an image processing section that divides the brightness data of pixels in the range of the metal strip plate into three components of R, G, and B values based on an image photographed by the camera, and detects a rolling abnormality of the surface based on the relationship of the brightness data of two components among the brightness data of the components.

[0015] Effects of the Invention

[0016] According to the present application, the precision of detection of a rolling abnormality of a plate can be improved compared to the past. The above problem, configuration, and effects will become clear from the following description of the embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a schematic diagram showing the configuration of a rolling apparatus having an abnormality detection device of an embodiment of the present application.

[0018] Figure 2 is a schematic diagram showing the configuration of an abnormality detection device of an embodiment of the present application.

[0019] Figure 3 is a diagram showing a case in which the range of a rolled material is extracted from an image in the abnormality detection device of the embodiment.

[0020] Figure 4 is a diagram showing an example of a brightness (RGB) histogram of pixels in the range of a plate for one image in the abnormality detection device of the embodiment.

[0021] Figure 5 is a diagram showing an example of the extraction determination process of the abnormal region in the abnormality detection device of the embodiment.

[0022] Figure 6 is a diagram showing an example of the extraction determination process of the abnormal region in the abnormality detection device of the embodiment.

[0023] Figure 7 is a diagram showing a flow of the calculation method of the abnormal region based on the reflected light region in the abnormality detection device of the embodiment.

[0024] Figure 8 is a diagram showing a flow of the calculation method of the abnormal region based on the reflected light region in the abnormality detection device of the embodiment.

[0025] Figure 9 is a diagram showing a flow of the calculation method of the abnormal region based on the reflected light region in the abnormality detection device of the embodiment.

[0026] Figure 10 is a flowchart of the abnormality determination process in the abnormality detection device of the embodiment. DETAILED DESCRIPTION

[0027] Using Figures 1-10 Embodiments of an abnormality detection device and an abnormality detection method of the present application will be described. In the drawings used in the present specification, the same or similar reference numerals are assigned to the same or corresponding constituent elements, and repetitive description will be omitted in some cases.

[0028] Further, the metal strip plate of the object material to be rolled in the present application is a strip plate of a metal material that can be generally rolled, and the kind thereof is not particularly limited, and a non-ferrous material such as aluminum or copper can be used in addition to a steel plate.

[0029] First, the overall configuration of a rolling apparatus including the abnormality detection device, and the configuration of the abnormality detection device will be described. Figure 1 and Figure 2 Embodiments of an abnormality detection device and an abnormality detection method of the present application will be described. In the drawings used in the present specification, the same or similar reference numerals are assigned to the same or corresponding constituent elements, and repetitive description will be omitted in some cases. Figure 1 is a diagram showing the configuration of the abnormality detection device of the present embodiment, Figure 2 is a diagram showing the configuration of the abnormality detection device.

[0030] Figure 1The rolling apparatus 100 shown is a finishing rolling apparatus for rolling a rolled material 1, and has an F1 stand 10, an F2 stand 20, an F3 stand 30, an F4 stand 40, an F5 stand 50, an F6 stand 60, an F7 stand 70, a camera 81, 82, a loop 65 for tension control, an image processing computer 90, a display device 95, and the like.

[0031] Among them, an abnormality detection device 101 that detects abnormalities of the surface of the rolled material 1 is constituted by the cameras 81, 82, the lighting device 67, and the image processing computer 90.

[0032] Further, with respect to the rolling apparatus 100, it is not limited to Figure 1 The form shown with seven stands, but it is not limited to this, and it can be one or more than one.

[0033] The F1 stand 10, the F2 stand 20, the F3 stand 30, the F4 stand 40, the F5 stand 50, the F6 stand 60, and the F7 stand 70 are rolling mills, each of which has an upper work roll and a lower work roll, and an upper backup roll and a lower backup roll that are supported by being brought into contact with the upper work roll and the lower work roll, respectively, a press-down cylinder 11, 21, 31, 41, 51, 61, 71 provided at the upper portion of the upper backup roll, a load detector 12, 22, 32, 42, 52, 62, 72. Further, it can be configured to have six stages in which an upper intermediate roll and a lower intermediate roll are respectively provided between each of the upper and lower work rolls and each of the upper and lower backup rolls. The roll configuration of the rolling mill is not limited to the form described above, and it can have at least the upper and lower work rolls.

[0034] The loop 65 is a roll for line tension control. This roll is provided so that its rotation axis extends in the width direction of the rolled material 1, and is provided between the F1 stand 10 and the F2 stand 20, between the F2 stand 20 and the F3 stand 30, between the F3 stand 30 and the F4 stand 40, between the F4 stand 40 and the F5 stand 50, between the F5 stand 50 and the F6 stand 60, and between the F6 stand 60 and the F7 stand 70, in a manner that enables the line tension to be changed by raising the rolled material 1 upward in the vertical direction or lowering the rolled material 1 downward. Further, the loop 65 can have a sheet shape meter function that detects the tension distribution in the sheet width direction.

[0035] The camera 81 is provided at a position where it can take an image of the rolled material 1 including between the exit side of the F4 stand 40 and the entry side of the F5 stand 50, and takes an image including the rolled material 1 in the form of a dynamic image (video) at appropriate intervals, for example, shorter than 0.1 seconds, from directly above or obliquely above the rolled material 1. The data of the image taken by the camera 81 is transmitted to the image processing computer 90 via a communication line 85.

[0036] The camera 82 is provided at a position capable of capturing an image including the rolled material 1 on the exit side of the F7 stand 70, and like the camera 81, captures an image including the rolled material 1 from directly above or obliquely above the rolled material 1 at intervals of, for example, less than 0.1 seconds. Data of the image captured by the camera 82 is also transmitted to the image processing computer 90 via the communication line 85.

[0037] A capturing process of capturing an image including the rolled material 1 by the cameras 81, 82 is executed.

[0038] Further, the case where the cameras are provided at both the exit side of the F4 stand 40 and the entry side of the F5 stand 50 and at the exit side of the F7 stand 70 is described, but the cameras can be provided at least at one place, and one or more than two can be provided, or can be provided between all stands or at the entry and exit sides of the rolling mill 100.

[0039] The lighting device 67 irradiates the rolled material 1, particularly the rolled material 1 in the range captured by the cameras 81, 82, and can be provided as a general lighting device appropriately arranged at the top of a rolling mill factory where the rolling mill 100 is installed or the like, but a dedicated lighting device can also be used.

[0040] The image processing computer 90 is a device constituted by a computer or the like that controls the operation of each machine in the rolling mill 100, and has an image processing section 92, a database 93, and the like.

[0041] The image processing section 92 is a section that separates the luminance data in the range of the rolled material 1 into three components of R, G, and B values from the image captured by the cameras 81, 82, and detects a rolling abnormality of the surface based on the relationship of the luminance data of two components among the luminance data of each component, and is the main body of execution of the image processing process.

[0042] The image processing section 92 has a detection region setting section 92A, a reflected light region setting section 92B, and an abnormality region setting section 92C. The detection region setting section 92A is a section that sets a detection region of the rolled material 1 that becomes a detection target of an abnormality from the captured image. The reflected light region setting section 92B is a section that sets a reflected light region of the irradiation of the illumination light from the lighting device 67 on the surface of the rolled material 1 from the captured image. The abnormality region setting section 92C is a section that detects an abnormal part of the surface from the detection region with a reference value as a threshold value, and sets a region after the reflected light region is removed as an abnormality region.

[0043] The detailed contents of the processing of the image processing section 92, or the detection region setting section 92A, the reflected light region setting section 92B, and the abnormal region setting section 92C in the image processing section 92 are described later.

[0044] The database 93 is a storage device that stores information of a threshold boundary that distinguishes a normal value from an abnormal value, which is previously calculated from a distribution of a plurality of plots in a two-dimensional graph composed of two components, of a difference in luminance of one component with respect to a difference in luminance of another component, of R (red) values, G (green) values, and B (blue) values possessed by pixels in a range of the rolled material 1 in an image in which the rolled material 1 is photographed, for example, stores information of a boundary line shown in Figure 5 or Figure 6 The information of the boundary line shown. It is appropriately composed of an SSD or an HDD, or the like.

[0045] The display device 95 is a display device such as a display or an acoustic device such as an alarm, for example, is a device for informing an operator of the occurrence of a plate shape abnormality and countermeasure work therefor when the plate shape abnormality is judged in the image processing computer 90. Therefore, a display is mostly used as the display device 95.

[0046] The operator confirms the presence or absence of a rolling abnormality such as a rolled plate extrusion phenomenon by visually observing the display screen of the display device 95 or each stand itself or between each stand in the mechanical work. For example, the operator confirms the display of the occurrence of the rolled plate extrusion phenomenon in the display device 95 and, in a case where an operation to be performed such as a bending correction, a leveling correction, a roll speed correction, a roll gap opening of a downstream stand, or the like is displayed, can avoid the transition from the state of the rolling abnormality to a further deteriorated state.

[0047] A form in which the occurrence of the abnormality of the rolled plate extrusion phenomenon is displayed to the operator on the display screen of the display device 95 and a signal for a correction or a stop for avoiding the rolling abnormality is transmitted to the rolling mill control device to automatically control the rolling abnormality avoidance operation, or a form in which various correction operations or a stop of the mechanical work are automatically controlled without being displayed on the display device 95 can be adopted.

[0048] Next, a specific example of the abnormality detection processing of judging the abnormality of the plate shape of the rolled material 1 in the present application is described. Figure 3 Next, a specific example of the abnormality detection processing of judging the abnormality of the plate shape of the rolled material 1 in the present application is described. Figure 3 is a diagram showing a case of extracting a range of the rolled material from an image, Figure 4 is a diagram showing an example of a luminance (RGB) histogram within a plate range with respect to one image, Figure 5 and Figure 6 is a diagram showing an example of an extraction judgment processing of an abnormal region, Figures 7-9 is a diagram showing a flow of a calculation method of an abnormal region based on a reflected light region.

[0049] First, the image including the rolled material 1 is acquired by the cameras 81, 82. The captured image is output to the image processing computer 90 via the communication line 85.

[0050] As for the image processing computer 90, in the image processing section 92, as shown in FIG. 6, the background and the region where the rolled material 1 exists are judged from the image of the rolled material 1 at the time of rolling by binarization processing, and the range 1A of the rolled material 1 is extracted from the luminance region above a certain threshold value. The abnormality detection processing is performed with respect to the extracted range 1A. Figure 3

[0051] Further, the extracted range 1A can include a region slightly inside or slightly outside the actually existing plate region, and does not necessarily have to coincide with the rolled material 1, but since there is a case where the position of the captured rolled material 1 is changed by making the loop 65 up and down, it is necessary to extract the range 1A in order to follow such a change.

[0052] Next, as for the image processing computer 90, in the image processing section 92, the luminance distribution of the range 1A of the rolled material 1 is found with respect to one selected image. Here, the image processing section 92 divides the luminance (RGB) data of all the pixels of the range 1A extracted in FIG. 5 with respect to one image captured by the cameras 81, 82 into three components of R value, G value, and B value, and finds the degree distribution (histogram) of the luminance of each of the R value, G value, and B value expressed in the number of measurement points. Here, R: red, G: green, B: blue, and each of RGB takes a value of 0-255. Figure 3 Figure 4 is an example showing the luminance (RGB) histogram of the plate range with respect to one image.

[0053] Next, the image processing section 92 takes a value dividing the area expressed by the integral value of each distribution into two parts (center position: 50%) as the luminance data of each RGB component of the plate judged to have no abnormality. In addition, one value can be selected within a tolerance range of ±20% from the center position 50% as the luminance data of each RGB component. Further, the tolerance range does not need to be "±20%", and can be appropriately changed.

[0054] In addition, instead of using the value dividing the area into two parts, it is also possible to set the luminance (the value of the highest frequency) that is the most in the number of measurement points in each distribution as the RGB luminance data of the plate judged to have no abnormality. It is also possible to select one value within a tolerance range of ±30% from the value of the highest frequency as the luminance data of each RGB component. Further, the tolerance range does not need to be "±30%", and can be appropriately changed.

[0055] ​​The value (position) used to find the brightness data can be in the form of a value specified by an operator before rolling the rolled material 1 to be subjected to anomaly detection, in the form of a value preset using the device, or in the form of a value appropriately learned as an optimum value using machine learning.

[0056] The image processing section 92 detects an anomaly based on the brightness data of one component, the brightness data of the other component, and the threshold boundary of the surface. Here, in a case where the detection result of an anomaly is obtained from the brightness data of two components that become one combination among the R value, the G value, and the B value and the threshold boundary obtained from them, and the detection result of an anomaly is obtained from the brightness data of two components that become another combination, the surface is detected as an anomaly.

[0057] For example, the image processing section 92 obtains the detection result of whether an anomaly is present or not from the brightness data of two components that become one combination (the brightness data of the B value and the brightness data of the G value) and the boundary obtained from them, as shown in FIG. 6A, and obtains the detection result of whether an anomaly is present or not from the brightness data of two components that become another combination (the brightness data of the B value and the brightness data of the R value) and the boundary obtained from them, as shown in FIG. 6B. Figure 5 Figure 6 More specifically, it is determined whether the plot of the brightness data is on the anomaly side compared to the boundary line.

[0058] Thereafter, the image processing section 92 detects the surface as an anomaly in a case where both of the detection results are anomalies.

[0059] Further, although not illustrated, the relationship between the brightness data of the G value and the brightness data of the R value can be made, but the image processing section 92 preferably uses the brightness data of the G value and the B value and the brightness data of one of the three combinations.

[0060] In addition, the case where two of the three combinations are used is described, but one of the three combinations can be used for determination. In addition, an anomaly can be detected in a case where one of the two combinations is equal to or higher than an arbitrary threshold boundary. All of the three combinations can also be used for determination. In a case where all of the three combinations are used, an anomaly can be detected in a case where one of them is equal to or higher than an arbitrary threshold boundary, in a case where two of them are equal to or higher than an arbitrary threshold boundary, or in a case where all of them are equal to or higher than an arbitrary threshold boundary. In these cases, as described above, the brightness data of the G value and the B value can be preferentially used.

[0061] For example, the image processing section 92 obtains the detection result of whether an anomaly is present or not from the brightness data of two components that become one combination (the brightness data of the B value and the brightness data of the G value) and the boundary obtained from them, as shown in FIG. 6A, and obtains the detection result of whether an anomaly is present or not from the brightness data of two components that become another combination (the brightness data of the B value and the brightness data of the R value) and the boundary obtained from them, as shown in FIG. 6B. Figure 5 Figure 6 ​​It is preferable to collect a plurality of normal / abnormal image data in advance for the arbitrary threshold boundary line shown, calculate an approximate formula that can separate them in advance, and store them in the database 93 in advance.

[0062] For example, an operator of the rolling mill 100 or an employee of the manufacturer of the abnormality detection device can make a judgment based on the result graph and draw an arbitrary threshold boundary line for separation. Alternatively, mathematical techniques, such as data clustering techniques (data classification techniques), can be used to classify the data into normal and abnormal. One such technique is to use a binary classification method based on the SVM (support vector machine) method to determine a linear equation so that the distance (difference) between normal and abnormal data is maximized.

[0063] Next, use Figures 7-9 The details of removing the reflected light area formed by the lighting device 67 will be described.

[0064] like Figure 7 As shown in FIG. 1 , the abnormal area is in the shape of the brightness distribution in the rolling direction (transportation direction). In contrast, in the image actually captured by the cameras 81 and 82, the abnormal area and the area of ​​the reflected light from the lighting device 67 are mixed (in the image). Figure 7 The image of the squeezed portion and reflected light area 1A1 in the width (longitudinal) direction of the middle plate is obtained by removing the reflected light area formed by the lighting device 67 from the squeezed portion and reflected light area 1A1.

[0065] Here, the lighting device 67 is close to white and may be close to yellow during squeezing. Therefore, it is desirable to perform binarization processing to extract and remove the light from the lighting device 67 from a brightness threshold close to white.

[0066] Then, in the image processing unit 92, the detection area setting unit 92A is used to set the above-mentioned Figure 5 as well as Figure 6 The area judged as abnormal by the method of etc. is set as the squeeze part and reflected light area 1A1. Figure 8 As shown, the reflected light region setting unit 92B extracts and deletes the relatively long reflected light region 1A2 in the plate width direction formed by the lighting device 67. Figure 9 As shown, only the true abnormal region 1A3 (relatively long region in the conveying direction) is extracted. Thereafter, the abnormal region setting unit 92C sets the detected true abnormal region 1A3 as the abnormal region, and the process transitions to the final detection process of detecting the presence or absence of rolling abnormality.

[0067] In addition, the description from the use of Figure 5The abnormal squeeze part and reflected light area 1A1 of the surface is detected in the order of removing the reflected light area 1A2 formed by the lighting device 67, but it is also possible to remove the reflected light area 1A2 formed by the lighting device 67 after the range 1A of the detection plate, and then use Figure 8 The order of processing the abnormal rolling area is set.

[0068] Thereafter, the image processing unit 92 preferably removes the abnormal area set by the reflected light area formed by the lighting device 67 or uses Figure 5 When the total area value of each pixel of the abnormal area of ​​the abnormal part of the surface detected by methods such as , or the value obtained by dividing the total area value by the area of ​​range 1A of the entire rolled material 1 is greater than a certain threshold value (ε), it is judged that a rolling abnormality has occurred in the rolled material 1.

[0069] Next, refer to Figure 10 An abnormality detection method for detecting abnormalities on the surface of a rolled material 1 rolled by a rolling mill according to this embodiment will be described. Figure 10 This is a flowchart of an abnormality determination process in the abnormality detection device of the embodiment.

[0070] First, if Figure 10 As shown, before actually rolling, the relationship between the brightness data (R, G, B) of the two components is obtained using past data (see Figure 5 、 Figure 6 ), and create a boundary line that separates normal / abnormal, and record it in the database 93 in advance (step S101).

[0071] When rolling is performed, first, images are acquired by the cameras 81 and 82 (step S102). This step S102 corresponds to an imaging step.

[0072] Next, the image processing unit 92 of the image processing computer 90 determines whether the rolled material 1 is present in the image captured in step S102 (step S103). If the rolled material 1 is present, the process proceeds to step S104. If the rolled material 1 is not present, the process proceeds to step S110.

[0073] After setting the plate detection area (step S104), the image processing unit 92 determines the abnormality candidate based on the relationship between the two brightness components (step S105). At this time, the image processing unit 92 preferably performs a process for removing the reflected light area formed by the lighting device 67 (step S106).

[0074] Thereafter, in the image processing section 92, final abnormality determination processing is executed (step S107), and it is determined whether or not a rolling abnormality such as a squeeze has occurred (step S108). When it is determined that an abnormality has occurred, the processing proceeds to step S109, and when it is determined that no abnormality has occurred, the processing proceeds to step S110. These steps S104 to S108 correspond to the image processing procedure.

[0075] In step S108, when it is determined that an abnormality has occurred, the image processing section 92 records an abnormality occurrence flag = 1 (step S109), and in step S103, when it is determined that the rolled material 1 is not present, or in step S108, when it is determined that no abnormality has occurred, the image processing section 92 records an abnormality occurrence flag = 0 (step S110), and the processing is started again at the next time.

[0076] In the image processing computer 90, when the abnormality occurrence flag 1 is recorded, the message is displayed to the display device 95. Alternatively, an intervention process with respect to each stand 10, 20, 30, 40, 50, 60, 70 can be automatically executed.

[0077] Next, the effects of the present embodiment will be described.

[0078] The abnormality detection device of the present embodiment described above is a device for detecting a rolling abnormality of a surface of a rolled material 1 rolled by a rolling mill, and includes a camera 81, 82 that photographs the rolled material 1 that is a detection target of an abnormality, and an image processing section 92 that divides the luminance data of the pixels within the range of the rolled material 1 into three components of R, G, and B values based on the image photographed by the camera 81, 82, and detects a rolling abnormality of the surface based on the relationship of the luminance data of two components among the luminance data of the respective components.

[0079] As such, the three primary colors of the color that uses the luminance distribution of the image close to the actual image are used, and the abnormality is determined using the luminance data of two components, and thus, compared to the past, the precision of the rolling abnormality of the rolled material 1 can be improved.

[0080] Further, if data of luminance is plotted for each of combinations of R and G values, G and B values, and R and B values, which are combinations of two kinds of components, it is seen that data groups of a sheet being normal and data groups of a sheet being abnormal are separated by a line as a boundary. Thus, the image processing section 92 is further provided with a database 93 that stores a threshold boundary for distinguishing between a normal value and an abnormal value, which is obtained in advance from a plurality of distributions plotted in a two-dimensional graph of two kinds of components, with one kind of component's luminance data plotted against another kind of component's luminance data, in an image captured by the imaging device 65. The image processing section 92 detects a rolling abnormality of the surface based on one kind of component's luminance data, another kind of component's luminance data, and the threshold boundary, thereby improving the accuracy of abnormality detection compared to a case where it is determined whether or not there is an abnormality using a threshold value determined for each kind of component.

[0081] Further, the image processing section 92 uses luminance differences of G and B values. Since the rolled material 1 is reddish, there is a case where the luminance of a normal portion and an abnormal portion hardly changes if R values are used, which is not suitable for determination. Thus, by using G and B values, it is possible to perform abnormality detection without taking into account the base color of the rolled material 1.

[0082] Further, the image processing section 92 detects a rolling abnormality when the detection result obtained from luminance data of two kinds of components that become one combination among R, G, and B values and a threshold boundary obtained therefrom is abnormal, and the detection result obtained from luminance data of two kinds of components that become another combination and a threshold boundary obtained therefrom is abnormal, thereby using two combinations of two kinds of components to determine that the surface is abnormal when an abnormality is detected in both, so that it becomes less likely to erroneously detect a normal portion as an abnormal portion, and it is possible to further improve the accuracy of abnormality detection.

[0083] Further, the image processing section 92 detects a rolling abnormality when the detection result obtained from luminance data of two kinds of components that become one combination among R, G, and B values and a threshold boundary obtained therefrom is abnormal, and the detection result obtained from luminance data of two kinds of components that become another combination and a threshold boundary obtained therefrom is abnormal, thereby using two combinations of two kinds of components to determine that the surface is abnormal when an abnormality is detected in both, so that it becomes less likely to erroneously detect a normal portion as an abnormal portion, and it is possible to further improve the accuracy of abnormality detection.

[0084] <Other>

[0085] Further, the present application is not limited to the above-described embodiments, and various modifications and applications can be made. The above-described embodiments are described in detail for the purpose of facilitating understanding of the present application, but are not limited to all the configurations described.

[0086] Reference Signs

[0087] 1 Rolled material (metal strip)

[0088] 1A Extracted range of the rolled material

[0089] 1A1 Pressing portion and reflected light region

[0090] 1A2 Reflected light region

[0091] 1A3 Abnormal region

[0092] 10F1 Stand

[0093] 11, 21, 31, 41, 51, 61, 71 Pressing cylinder

[0094] 12, 22, 32, 42, 52, 62, 72 Load detector 20F2 Stand

[0095] 30F3 Stand

[0096] 40F4 Stand

[0097] 50F5 Stand

[0098] 60F6 Stand

[0099] 61 Pressing cylinder

[0100] 65 Loop for tension control

[0101] 67 Illumination device

[0102] 70F7 Stand

[0103] 71 Pressing cylinder

[0104] 81, 82 Camera

[0105] 85 Communication line

[0106] 90 Image processing computer

[0107] 92 Image processing section

[0108] 92A Detection region setting section

[0109] 92B Reflected light region setting section

[0110] 92C Abnormal region setting section

[0111] 93 Database

[0112] 95 Display device

[0113] 100 Rolling apparatus

[0114] 101 Anomaly detection apparatus.

Claims

1. An abnormality detection device for detecting rolling abnormalities on the surface of a metal strip rolled by a rolling mill, the abnormality detection device comprising: a camera for photographing the metal strip plate as an abnormality detection target; an image processing unit that separates brightness data of pixels within the metal strip into three components, namely, R value, G value, and B value, based on the image captured by the camera, and detects rolling abnormalities on the surface based on a relationship between the brightness data of two components among the brightness data of each component; and A database having threshold boundaries for distinguishing normal values ​​from abnormal values ​​determined in advance based on a specific distribution, wherein a plurality of brightness data of one component among R value, G value, and B value in an image of the metal strip are plotted against brightness data of another component in a two-dimensional graph consisting of two components, When the image processing unit detects rolling abnormality on the surface based on the brightness data of one component, the brightness data of the other component, and the threshold boundary, the image processing unit detects the rolling abnormality when the detection results obtained based on the brightness data of two components forming a combination among the R value, G value, and B value and the threshold boundary calculated therefrom are abnormal, and when the detection results obtained based on the brightness data of two components forming another combination and the threshold boundary calculated therefrom are abnormal.

2. The abnormality detection device according to claim 1, characterized in that The image processing unit uses the brightness data of the G value and the brightness data of the B value.

3. The abnormality detection device according to claim 1 or 2, characterized in that: It also includes a lighting device for illuminating the metal strip, The image processing unit sets the detection area of ​​the metal strip based on the latest image, sets the reflected light area of ​​the illumination light from the lighting device irradiated on the surface of the metal strip based on the latest image, and resets the area after removing the reflected light area from the detection area as the detection area, and uses the brightness data of the pixels in the newly obtained detection area to detect the rolling abnormality.

4. A method for detecting anomaly on the surface of a metal strip rolled by a rolling mill, the method comprising: a photographing step of photographing the metal strip plate as a detection target for abnormality; as well as an image processing step of separating brightness data of pixels within the metal strip into three components, namely, R value, G value, and B value, based on the image captured in the capturing step, and detecting rolling abnormalities on the surface based on a relationship between the brightness data of two components among the brightness data of each component; The threshold boundary for distinguishing normal values ​​from abnormal values ​​is determined based on a specific distribution in which the brightness data of one component among the R value, G value, and B value in the image of the metal strip is plotted against the brightness data of the other component in a two-dimensional graph consisting of two components. In the image processing step, when detecting rolling abnormality of the surface based on the brightness data of the one component, the brightness data of the other component, and the threshold boundary, if the detection result obtained based on the brightness data of two components forming a combination among the R value, G value, and B value and the threshold boundary calculated therefrom is abnormal, and if the detection result obtained based on the brightness data of two components forming another combination and the threshold boundary calculated therefrom is abnormal, the rolling abnormality is detected.

5. The anomaly detection method according to claim 4, characterized in that: The brightness data of the G value and the brightness data of the B value are used in the image processing step.

6. The abnormality detection method according to claim 4 or 5, characterized in that: In the image processing step, the detection area of ​​the metal strip is set according to the latest image, the reflected light area of ​​the illumination light irradiated on the surface of the metal strip is set according to the latest image, the area after removing the reflected light area from the detection area is reset as the detection area, and the brightness data of the pixels in the newly obtained detection area is used to detect the rolling abnormality.

7. An abnormality detection device for detecting rolling abnormalities on the surface of a metal strip rolled by a rolling mill, the abnormality detection device comprising: a camera for photographing the metal strip plate as an abnormality detection target; an image processing unit that separates brightness data of pixels within the metal strip into three components, namely, R value, G value, and B value, based on the image captured by the camera, and detects rolling abnormalities on the surface based on a relationship between the brightness data of two components among the brightness data of each component; and a lighting device for illuminating the metal strip, The image processing unit sets the detection area of ​​the metal strip based on the latest image, sets the reflected light area of ​​the illumination light from the lighting device irradiated on the surface of the metal strip based on the latest image, and resets the area after removing the reflected light area from the detection area as the detection area, and uses the brightness data of the pixels in the newly obtained detection area to detect the rolling abnormality.

8. A method for detecting anomaly on the surface of a metal strip rolled by a rolling mill, the method comprising: a photographing step of photographing the metal strip plate as a detection target for abnormality; as well as an image processing step of separating brightness data of pixels within the metal strip into three components, namely, R value, G value, and B value, based on the image captured in the capturing step, and detecting rolling abnormalities on the surface based on a relationship between the brightness data of two components among the brightness data of each component; In the image processing step, the detection area of ​​the metal strip is set according to the latest image, the reflected light area of ​​the illumination light irradiated on the surface of the metal strip is set according to the latest image, the area after removing the reflected light area from the detection area is reset as the detection area, and the brightness data of the pixels in the newly obtained detection area is used to detect the rolling abnormality.

Images