Edge detection method, edge detection device and leveling control method

Abstract

To provide a technology capable of accurately detecting an edge of a sheet-like plate being conveyed without depending on a disturbance influence such as steam or a scale.SOLUTION: An edge detection method for detecting an edge of a sheet-like plate from an image captured consecutively in a longitudinal direction of the plate being conveyed. This method includes: acquiring a luminance differential value in a width direction of the plate from the captured image; determining coordinates at which a differential intensity integrated value obtained by integrating the luminance differential value in the longitudinal direction of the plate becomes maximum; and detecting edge coordinates of the plate on the basis of the coordinates.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present invention relates to an edge detection method and an edge detection device for detecting an edge of a sheet-like plate material to be conveyed, and a leveling control method.

Background Art

[0002] In the hot rolling process, troubles called end squeezing, in which the trailing end portion of the hot rolled steel sheet collides with the side guide and folds, frequently occur due to the meandering of the hot rolled steel sheet or the bending of the hot rolled steel sheet during the passage of the finishing rolling mill. For this reason, the operator visually checks the meandering and bending of the hot rolled steel sheet during the passage, and suppresses the occurrence of such meandering and bending of the hot rolled steel sheet during the passage by manually adjusting the left-right opening difference of the finishing rolling mill, called leveling. However, in order to perform a leveling operation on the hot rolled steel sheet during the passage of the finishing rolling mill, a high level of skill is required of the operator, and there are also differences in the degree of skill proficiency among operators, so it has become difficult to eliminate end squeezing due to the meandering and bending of the hot rolled steel sheet.

[0003] Therefore, recently, control technologies for automatically performing leveling according to the meandering and bending of the hot rolled steel sheet during the passage have been developed. In this automatic leveling control technology, a camera installed between the stands of the finishing rolling mill continuously photographs the hot rolled steel sheet during the passage, and the leveling is automatically operated based on the amount of meandering and the amount of bending of the hot rolled steel sheet calculated from the photographed image.

[0004] Tail-end narrowing is a phenomenon that occurs prominently mainly when passing a hot-rolled steel sheet with a thin thickness through the final stand with a high sheet speed. In order to perform leveling automatic control most effectively at the final stand, it is common to adopt a line sensor with high resolution and high responsiveness. However, in the finishing stand of the hot rolling process, a large amount of steam, scale, etc. are generated during the passage of the hot-rolled steel sheet, and it is assumed that these will cover the field of view of the sensor. The field of view of the line sensor is one-dimensional. When steam or scale appears in the field of view at the moment of shooting, the steel sheet cannot be photographed at all. Therefore, when such a disturbance occurs, abnormal shooting will occur, and there is a risk of performing an unexpected leveling operation.

[0005] Therefore, by using a two-dimensional imaging device such as a two-dimensional CCD camera (hereinafter referred to as an area camera), a method of stably measuring the meandering and bending of a steel sheet even in a harsh environment such as the hot rolling process has been adopted.

[0006] In the measurement of the meandering and bending of a hot-rolled steel sheet using an area camera, it is possible to take the imaging field of view two-dimensionally. Therefore, even when steam, scale, etc. occur locally at the shooting timing, if there is a range in the shooting range that is not covered by steam, scale, etc., there is a merit that it is possible to detect the edge position of the steel sheet.

[0007] Utilizing this merit, in Patent Document 1, for the photographed image by an area camera, the edge position of the steel sheet is calculated for each scanning line in the sheet width direction, and an approximate straight line of the steel sheet edge is calculated from the least squares method applied to each detected edge position, thereby minimizing the influence of extreme measurement errors caused by disturbances such as steam and scale. Also, in Patent Document 2, a method of specifying the edge is proposed by obtaining a straight line in the synthetic differential image obtained by synthesizing the differential image obtained from the differential intensity in the photographed image with a plurality of temporally continuous differential images and finding the straight line where the sum of the differential intensities of the pixels existing linearly is the maximum.

Prior Art Documents

Patent Documents

[0008] Patent Document 1 Japanese Patent Application Laid-Open No. 2004-141956 Patent Document 2 Japanese Patent Application Laid-Open No. 2012-26767 Summary of the Invention Problems to be Solved by the Invention

[0009] However, in the prior art as described above, there is a problem that when steam or scale is reflected in a wide range of the captured image and most of the steel plate cannot be observed, edge detection cannot be appropriately performed.

[0010] Therefore, the present invention provides an edge detection method and an edge detection device that can accurately detect the edge of a conveyed sheet-like plate material without being affected by disturbances such as steam and scale, and a leveling control method for performing leveling control of a rolling device that rolls a conveyed sheet-like plate material. Means for Solving the Problems

[0011] In order to solve the above problems, the present invention provides the following [1] to

[14] .

[0012] [1] An edge detection method for detecting an edge of a conveyed sheet-like plate material from an image continuously captured in the longitudinal direction of the plate material, comprising: obtaining a luminance differential value in the width direction of the plate material from the captured image; finding a coordinate at which a differential intensity integration value obtained by integrating the luminance differential value in the longitudinal direction of the plate material is maximized, and detecting an edge coordinate of the plate material based on the coordinate.

[0013] [2] The edge detection method according to [1], wherein the differential intensity integration value is obtained from the total length in the longitudinal direction of each image continuously captured over the entire length of the plate material.

[0014] [3] Divide the image of the sheet material into a plurality of parts in the longitudinal direction of the sheet material, obtain the luminance differential value in the width direction in each divided region, obtain the coordinate at which the integrated differential intensity value obtained by integrating the luminance differential values of each divided region in the longitudinal direction is maximized, and detect the edge of each divided region from the coordinate. The edge detection method according to [1].

[0015] [4] The number of divisions of the image of the sheet material is 3 to 10. The edge detection method according to [3].

[0016] [5] Detect the width direction coordinate at which the integrated differential intensity value obtained from the total length in the longitudinal direction of the captured image is maximized as the detection reference coordinate, and perform edge detection of each divided region with reference to the detection reference coordinate. The edge detection method according to [3].

[0017] [6] Determine the coordinate at which the differential integration value in each divided region is maximized within the range of ±5 to 40 mm of the detection reference coordinate as the edge coordinate in each divided region. The edge detection method according to [5].

[0018] [7] Determine whether the integrated differential intensity value is less than the threshold value. If it is less than the threshold value, determine it as a value with low reliability, and use the integrated differential intensity value of the image captured one frame before. The edge detection method according to any one of [1] to [6].

[0019] [8] If the coordinate of the detected edge is different from the edge coordinate of the image captured one frame before by a value of 30 to 70 mm or more within the range, determine it as a value with low reliability, and use the edge coordinate of the image captured one frame before. The edge detection method according to any one of [1] to [6].

[0020] [9] An edge detection device for detecting the edge of a sheet-shaped sheet material continuously imaged in the longitudinal direction, A luminance differential value acquisition unit that acquires the luminance differential value in the width direction of the sheet material from the captured image, A differential intensity integrated value calculation unit that calculates the integrated differential intensity value obtained by integrating the luminance differential values in the longitudinal direction of the sheet material, An edge coordinate calculation unit that obtains the coordinates at which the differential intensity integrated value is maximized; An edge detection device having the above, and based on the coordinates, the edge coordinates of the sheet material are detected.

[0021]

[10] Further comprising an image division unit that divides the captured image into a plurality of parts in the longitudinal direction of the sheet material, The luminance differential value acquisition unit acquires the luminance differential value in the width direction in each divided region divided by the division unit, the differential intensity integrated value calculation unit obtains the differential intensity integrated value of each divided region, and the edge coordinate calculation unit obtains the coordinates at which the partial intensity integrated value in each divided region is maximized. The edge detection device according to [9].

[0022]

[11] The edge coordinate calculation unit determines, as the edge coordinates in each divided region, the coordinates at which the differential integrated value in each divided region is maximized within a range of ±5 to 40 mm of the detection reference coordinates. The edge detection device according to

[10] .

[0023]

[12] Further comprising a differential intensity integrated value determination unit that compares the differential intensity integrated value with a preset threshold value and determines that the value is unreliable when the differential intensity integrated value is less than the threshold value. The edge detection device according to any one of [9] to

[11] .

[0024]

[13] If the coordinates of the detected edge are different by a value greater than or equal to a certain value within a range of ±30 to 70 mm from the edge coordinates of the image captured one time before, it is determined that the value is of low reliability, and the edge coordinate determination unit that uses the edge coordinates of the image captured one time before. The edge detection device according to any one of [9] to

[11] .

[0025]

[14] A leveling control method for performing leveling control of a rolling device that rolls a conveyed sheet-like sheet material, A step of detecting the edge coordinates of the sheet material from an image continuously captured in the longitudinal direction of the conveyed sheet-like sheet material, A step of calculating the amount of meandering and the amount of bending of the plate material based on the detected edge coordinates; A step of applying a leveling control signal to the rolling mill based on the calculated amount of meandering and the amount of bending; having The step of detecting the edge coordinates of the plate material is From the captured image, obtain the luminance differential value in the width direction of the plate material, Determine the coordinates at which the integrated differential intensity value obtained by integrating the luminance differential value in the longitudinal direction of the plate material is maximized, and based on the coordinates, detect the edge coordinates of the plate material. A leveling control method.

Effect of the Invention

[0026] According to the present invention, since a method of integrating the luminance differential values of the entire length in the longitudinal direction of a sheet-like plate material and determining the width direction coordinates where the luminance change is the largest in the captured image is used, even when transporting the sheet-like plate material in an environment where a lot of steam, scale, etc. appear, it is possible to accurately detect the edge of the plate material without being affected by such disturbances.

Brief Description of the Drawings

[0027]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

BEST MODE FOR CARRYING OUT THE INVENTION

[0028] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the present embodiment, as a method for detecting the edge of a sheet-shaped plate material to be conveyed, in a hot rolling process of hot rolling a steel material as a material to be rolled, an edge detection method of a hot-rolled steel sheet when finish rolling in a stand of a continuous finishing mill will be described as an example.

[0029] FIG. 1 is a schematic diagram showing a continuous finishing mill facility provided with an area camera which is a two-dimensional imaging device for detecting the edge of a hot-rolled steel sheet, and FIG. 2 is a configuration diagram showing a control system for detecting the edge of a hot-rolled steel sheet from an image taken using the area camera and performing leveling control of the finishing mill.

[0030] As shown in FIG. 1, the continuous finishing mill facility 10 is for finish rolling the hot-rolled steel sheet 1, and the hot-rolled steel sheet 1 is sequentially passed through a plurality of finishing mill stands 2 and rolled to a desired thickness. The area camera 3 continuously images the hot-rolled steel sheet 1 passing between the finishing mill stands 2 in the longitudinal direction.

[0031] As shown in FIG. 2, the control system 20 includes an arithmetic unit (PC) 4 that detects the edge of the hot-rolled steel sheet 1 and calculates the amount of meandering and the amount of bending of the hot-rolled steel sheet, and a control unit 5 that gives a leveling control signal to the finishing mill stand 2 based on the calculation result of the arithmetic unit 4. The image taken by the area camera 3 is transmitted to the arithmetic unit 4 through the optical cable 8.

[0032] The arithmetic unit 4 includes an edge detection device 6 that detects the coordinates of the edge of the hot-rolled steel sheet 1 based on the image captured by the area camera 3, and a serpentine amount and bend amount calculation device 7 that calculates the serpentine amount and bend amount of the hot-rolled steel sheet 1 based on the coordinates of the edge of the hot-rolled steel sheet 1 detected by the edge detection device 6. Then, the control unit 5 gives a leveling control signal to the finishing rolling stand 2 based on the serpentine amount and bend amount of the hot-rolled steel sheet 1 calculated by the serpentine amount and bend amount calculation device 7.

[0033] As shown in FIG. 3, the edge detection device 6 includes a luminance distribution acquisition unit 11, an image segmentation unit 12, a differential intensity integrated value calculation unit 13, an edge coordinate calculation unit 14, a differential intensity integrated value determination unit 15, and an edge coordinate determination unit 16.

[0034] The luminance distribution acquisition unit 11 acquires the luminance differential value in the width direction (x direction) of the hot-rolled steel sheet 1 from the image captured by the area camera 3. The "luminance differential value" is the absolute value of the width direction differential value obtained by differentiating the luminance of the captured image in the width direction. The absolute value is used because the polarities are different at both edges of the hot-rolled steel sheet. The image segmentation unit 12 divides the image captured by the area camera 3 into 3 to 10 segments in the longitudinal direction (y direction) of the hot-rolled steel sheet 1. The differential intensity integrated value calculation unit 13 calculates the differential intensity integrated value. The "differential intensity integrated value" is the value obtained by integrating the luminance differential value in the longitudinal direction (y direction) of the hot-rolled steel sheet 1. The edge coordinate calculation unit 14 obtains the coordinates at which the differential intensity integrated value is maximum. Based on these coordinates at which the differential intensity integrated value is maximum, the edge coordinates of the hot-rolled steel sheet 1 are detected. The differential intensity integrated value determination unit 15 compares the differential intensity integrated value with a preset threshold value, and determines that the value is unreliable when the differential intensity integrated value is less than the threshold value. The edge coordinate determination unit 16 compares the obtained edge coordinates with the edge coordinates obtained from the previous image, and determines that the value has low reliability if the difference is large, for example, different by a value within the range of ±30 to 70 mm or more.

[0035] Next, the edge detection method performed by the edge detection device 6 will be described. FIG. 4 is a flowchart showing the edge detection method.

[0036] First, from the images continuously captured in the longitudinal direction of the hot-rolled steel sheet 1 conveyed by the area camera 3, the luminance distribution acquisition unit 11 acquires the luminance differential value in the width direction of the hot-rolled steel sheet 1 (step ST1). At this time, the resolution of the area camera 3 is important. It is desirable that the resolution of the area camera 3 is within a range that can sufficiently compensate for the movement amount of the hot-rolled steel sheet 1 in the width direction during the imaging period of the area camera 3. In the embodiment described later, assuming that the movement amount in the width direction per imaging frame is 10 mm, the resolution is set to 0.7 mm / pixel, which is 1 / 10 or less of that. Also, regarding the imaging period of the area camera 3, in order to continuously image the entire length of the hot-rolled steel sheet 1 without interruption, it is desirable to determine an imaging period that satisfies the condition of "imaging area ≥ steel sheet speed × imaging period".

[0037] Next, the image division unit 12 divides the image captured by the area camera 3 into a plurality of regions in the longitudinal direction (step ST2). The number of divisions at this time may be 3 to 10, and it is desirable to be about 3 to 5 divisions, as described later.

[0038] Next, the differential intensity integration value calculation unit 13 calculates the differential intensity integration value over the entire length of the hot-rolled steel sheet 1 in the longitudinal direction (step ST3). That is, the luminance differential values of the respective images continuously captured over the entire length of the hot-rolled steel sheet 1 by the area camera 3 are integrated in the longitudinal direction of the hot-rolled steel sheet 1.

[0039] Conventionally, when extracting the edge of a steel sheet from a captured image, for example, as shown in Patent Document 1, a method of obtaining the luminance differential value over a plurality of scanning lines in the width direction of the captured image and determining the point where the luminance differential value is maximum as the edge coordinate has been common. However, with such a method, when steam or scale is reflected in a wide range of the captured image and most of the steel sheet cannot be observed, it is not possible to appropriately detect the edge.

[0040] In contrast, this embodiment is a method for identifying the width-direction coordinate with the most significant luminance change in the captured image by obtaining a differential intensity integration value obtained by integrating the luminance differential values of the captured image in the longitudinal direction. That is, by obtaining the width-direction coordinate of the point where the differential intensity integration value is maximized, the width-direction coordinate with the largest luminance change in the captured image can be identified, and the edge of the steel plate can be detected. Therefore, even when the edge portion of the steel plate is hidden due to disturbance, it is possible to accurately detect the edge of the steel plate.

[0041] More specifically, taking the x coordinate in the width direction and the y coordinate in the longitudinal direction, and setting the luminance of the pixel at the coordinate (x, y) as Lum (x,y) When this is done, the luminance differential value in step ST1 and the differential intensity integration value in step ST3 are respectively represented by the following equations (1) and (2), and the x coordinate at which the differential intensity integration value Lum_sum x in equation (2) is maximized can be used as the edge coordinate.

Equation

[0042] In this way, the width-direction coordinate of the point where the differential intensity integration value is maximized may be used as the edge coordinate. However, in order to more accurately detect the edge coordinate, the following steps may be added to determine the edge coordinate.

[0043]

[0044] For example, it can be mentioned that the width-direction coordinate at which the differential intensity integration value is maximized for each divided region divided in the longitudinal direction is used as the edge coordinate. Also, a threshold value of the differential intensity integration value for the entire longitudinal length is set, and when it is less than the threshold value, it is determined as a value with low reliability, and edge detection is performed based on the image captured one frame before, or the width-direction coordinate at which the differential intensity integration value for the entire longitudinal length is maximized is used as the detection reference coordinate in the edge detection of each divided region, etc. As shown in FIG. 4, next to step ST3, the differential intensity integration value determination unit 14 compares the differential intensity integration value with a preset threshold value (step ST4). When the differential intensity integration value is equal to or greater than the threshold value, the differential intensity integration value is adopted (step ST5). When the differential intensity integration value is less than the threshold value, the differential intensity integration value of the previous image is adopted (step ST6). Then, the point at which the differential intensity integration value adopted in step ST5 and step ST6 is maximized is set as the detection reference coordinate.

[0045] Here, the "detection reference coordinate" refers to the width-direction coordinate at which the differential intensity integration value obtained from the total length in the longitudinal direction of the captured image is maximized. As will be described later, this is used as the reference for edge detection of the hot-rolled steel sheet in each divided region.

[0046] The reason for performing the threshold determination of the differential intensity integration value as in step ST4 is that in some cases, the entire image may be completely covered by noise, and there may be an image in which it is difficult to visually confirm the edge position of the hot-rolled steel sheet. Therefore, when the differential intensity integration value is less than the threshold value, it is determined that the value is unreliable. Specifically, such determination is performed for the edges on both the left and right sides of the hot-rolled steel sheet 1. When the differential intensity integration value of any one of the edges is less than the threshold value, it is determined in step ST6 that the value is unreliable, and the detection reference coordinate obtained from the differential intensity integration value of the image captured one frame before is set as the detection reference coordinate of the current image. When the differential intensity integration value in the image captured one frame before is also unreliable, the detection reference coordinate obtained from the differential intensity integration value of the image captured one frame before that is further used. That is, the previous images are traced back until an image with a reliable differential intensity integration value is obtained. It is desirable that the threshold value in step ST4 be a value equal to or greater than "the number of pixels in the longitudinal direction of the captured image × 2".

[0047] Next, a differential intensity integrated value is calculated for each divided region divided in the longitudinal direction in step ST2 (step ST7). If only the differential intensity integrated value is calculated over the entire length of the captured image and the coordinate at which the value is maximized is taken as the edge coordinate, only one point of the edge coordinate can be determined. When there is a bend in the steel plate in the captured image or when the steel plate is obliquely within the field of view, it is difficult to appropriately identify the edge position. Therefore, as in step ST7, the differential intensity integrated value is calculated for each divided region obtained in step ST2. The number of divisions at this time may be 3 to 10 as described above, and from the viewpoint of performing the calculation with a load considering the response time to the leveling control as well, it is desirable to set it to about 3 to 5 divisions. The calculation of the differential intensity integrated value for each divided region at this time is performed on each divided screen with the same logic as step ST3 based on the detection reference coordinate.

[0048] Next, a point at which the differential intensity integrated value becomes the maximum value is obtained within a predetermined range before and after the detection reference coordinate of each divided region, preferably within a range of ±5 to 40 mm (step ST8).

[0049] If the x coordinate at which the differential intensity integrated value is simply maximized for each divided region is taken as the edge coordinate, when there are extremely large luminance changes on the scanning line due to steam, shadows, reflected light, etc., there is a risk of detecting a position completely different from the edge of the hot-rolled steel plate as the edge coordinate.

[0050] Also, in many cases, in order to avoid such false detections, a method is used in which the location where the steel plate edge exists is estimated in advance, and the point at which the luminance change is maximized near the estimated coordinate is specified as the edge coordinate. In this case, often a method is used in which the captured image is binarized using Otsu's discriminant formula or the like, and the steel plate edge is searched near the coordinate of the black-and-white boundary line of the binarized image. However, in such a method, when a local dark part is formed in the captured steel plate image, or when the temperature is locally low and dark only at the end of the steel plate, etc., there are problems such that the image cannot be appropriately binarized and the edge coordinate cannot be appropriately estimated.

[0051] Therefore, in step ST8, edge detection is performed based on the detection reference coordinate, which is the width direction coordinate at which the integrated differential intensity value obtained from the full longitudinal length of the captured image is maximized. As a result, in each divided region, the point where the luminance change is most concentrated can be identified, so that the edge coordinates can be estimated with higher accuracy. As the search range for the edge, it is desirable to determine it based on the maximum value of the width direction movement amount of the hot-rolled steel sheet in the imaging cycle, and it is preferable to determine the coordinate at which the integrated differential intensity value in each divided region is maximized within the range of ±5 to 40 mm of the detection reference coordinate as the edge coordinate in each divided region.

[0052] Next, it is determined whether the coordinates of the point obtained in step ST8 differ from the previous edge coordinates by a value greater than or equal to a certain value within the range of ±30 to 70 mm (step ST9). If they do not differ by a value greater than or equal to a certain value within the range of ±30 to 70 mm, the point obtained in step ST8 is determined as the edge coordinate (step ST10). If they differ by a value greater than or equal to a certain value within the range of ±30 to 70 mm, the point obtained in step ST8 is excluded as a value with low reliability, and the point obtained based on the detection reference coordinate of the previous image is used as the edge coordinate of the current image (step ST11). Specifically, when the detected edge coordinate differs from the edge coordinate on either the left or right side of the image captured one before by a value greater than or equal to a certain value within the range of ±30 to 70 mm, the different edge coordinate is determined as having low reliability, and the edge coordinate detected by the image captured one before is used as the edge image of the current image. The determination threshold value in step ST9 is preferably determined based on the maximum value of the width direction movement amount of the hot-rolled steel sheet during the imaging cycle, and is set to a value within the range of ±30 to 70 mm as described above. In step ST11, if the detection reference coordinate in the previous image is also unreliable, similar to step ST6, the detection reference coordinate of the previous image before that is used, and if it is still unreliable, the previous image is traced back until an image with a reliable detection reference coordinate is obtained.

[0053] As described above, the edge coordinates are calculated for each divided region, and finally, the average of the individual edge coordinates thus calculated is used as the representative value of the edge coordinates. Of course, at this time, the points excluded as having low reliability in step ST11 are excluded from the averaging target.

[0054] After the edge of the hot-rolled steel sheet 1 is detected by the edge detection device 6 in this way, the meandering amount and the bending amount of the hot-rolled steel sheet 1 are calculated by the meandering amount and bending amount calculation device 7 based on the detected edge coordinates. Then, the control unit 5 gives a leveling control signal to the finishing rolling stand 2 based on the meandering amount and the bending amount of the hot-rolled steel sheet 1 calculated by the meandering amount and bending amount calculation device 7, and performs leveling control of the finishing rolling mill. Here, the meandering amount refers to the difference between the center coordinates of both edges and the mill center coordinate, and the bending amount refers to the difference between the center coordinates of both edges at the tip portion and the center coordinates of both edges at the steady portion. Note that the difference between the center coordinates of both edges at the trailing end portion and the center coordinate of the steady portion is defined as the trailing end bending amount.

[0055] According to the present embodiment, since the method of integrating the luminance differential values of the entire length in the longitudinal direction of the hot-rolled steel sheet and determining the width direction coordinate where the luminance change is the largest in the captured image is used, even if a part of the captured hot-rolled steel sheet is hidden by disturbances in an environment where a lot of steam, scale, etc. appear as in the hot rolling process, it is possible to accurately detect the edge of the hot-rolled steel sheet without being affected by these disturbances. In addition, the longitudinal direction information obtained from the two-dimensional image of the area camera, which is a two-dimensional imaging device, can be fully utilized.

[0056] Further, by calculating the differential intensity integration value in each divided region divided in the longitudinal direction and setting the coordinates where these values are the largest as the edge coordinates, a plurality of edge coordinates can be obtained. When there is a bend in the steel plate in the captured image or when the steel plate is obliquely within the field of view, the edge position can be appropriately specified. At this time, by using the detection reference coordinates as the reference for the width direction coordinates in each divided region, the point where the luminance change is most concentrated can be specified in each divided region, so that the edge coordinates can be estimated with higher accuracy.

[0057] Furthermore, it is also possible to determine the reliability of the detected steel plate edge by comparing the edge detection results based on the edge detection results up to the previous measurement image. That is, in this way, by effectively utilizing the steel plate longitudinal direction information and the information before and after continuous imaging, edge detection with higher accuracy than the conventional method becomes possible.

[0058] As described above, the embodiments of the present invention have been described, but these are merely examples and should be considered as non-limiting. The above embodiments may be omitted, substituted, or changed in various forms without departing from the gist of the present invention.

[0059] For example, in the above embodiment, the case of detecting the edge of a hot-rolled steel plate continuously conveyed in the hot rolling process has been described as an example, but other steel plates continuously conveyed are also acceptable, not limited to hot-rolled steel plates. Further, the present invention is applicable not only to steel plates but also to sheet-shaped plate materials continuously conveyed.

Example

[0060] Next, an example of the present invention will be described. Here, a total of 1425 hot-rolled steel plates rolled in the hot rolling process over two days were imaged with an area camera, which is a two-dimensional imaging device installed between stands in the front stage of rolling in a continuous finishing rolling mill. Edge detection was performed based on the above embodiment from the captured images, and the width of the hot-rolled steel plate was measured from the results. FIG. 5 is a diagram showing the result of comparing the width of each steel plate obtained from the result of edge detection in such an embodiment with the actual width of each steel plate visually grasped in pixel units from the captured image. FIG. 5(a) shows the result at a position 600 mm from the tip of the steel plate, and FIG. 5(b) shows the result at a position 5000 mm from the tip of the steel plate. The vertical axis in FIGS. 5(a) and 5(b) shows the measurement error of the steel plate width obtained from the result of edge detection with respect to the actual steel plate width.

[0061] As shown in FIG. 5, more than 99% is within a measurement error of ±10 mm of the steel plate width, and it was confirmed that the steel plate edge detection according to the present invention is possible without being affected by disturbances. Although there are some points where extremely large measurement errors are observed, as shown in FIG. 6, the entire image was hidden by steam, and there were images where it was difficult to detect the edge even visually. For such difficult-to-detect images, physical measures such as blowing the steam away with a blower may be taken.

[0062] Next, one of the steel plates for which the above width measurement was performed was taken out, and edge detection of the entire length of one steel plate was performed from the image captured based on the above embodiment, and the width measurement was performed. The result is shown in FIG. 7. As shown in this figure, it was confirmed that the width can be stably measured over the entire length of the steel plate. This is because the edge of the hot-rolled steel plate can be accurately detected without being affected by disturbances by the edge detection method of the above embodiment, and furthermore, no abnormal values are calculated by performing reliability determination of the detection result.

Explanation of Signs

[0063] 1 Hot-rolled steel plate 2 Finishing rolling stand 3 Area camera (two-dimensional imaging device) 4 Computing device (PC) 5 Control unit 6 Edge detection device 7 Snakiness amount and turning amount calculation device 11 Luminance differential value acquisition unit 12 Image segmentation unit 13 Differential intensity integrated value calculation unit 14 Edge coordinate calculation unit 15 Differential intensity integrated value determination unit 16 Edge coordinate determination unit

Claims

1. An edge detection method for detecting an edge of a sheet-shaped plate material continuously imaged in the longitudinal direction of the plate material to be conveyed, comprising: obtaining a luminance differential value in the width direction of the plate material from the captured image; obtaining a coordinate at which a differential intensity integration value obtained by integrating the luminance differential value in the longitudinal direction of the plate material is maximized, and detecting an edge coordinate of the plate material based on the coordinate.

2. The edge detection method according to claim 1, wherein the differential intensity integration value is obtained from the total length in the longitudinal direction of each image continuously captured over the entire length of the plate material.

3. The edge detection method according to claim 1, wherein the image of the plate material is divided into a plurality of parts in the longitudinal direction of the plate material, a luminance differential value in the width direction is obtained in each divided region, and a coordinate at which a differential intensity integration value obtained by integrating the luminance differential value of each divided region in the longitudinal direction is maximized is obtained, and an edge of each divided region is detected from the coordinate.

4. The edge detection method according to claim 3, wherein the number of divisions of the image of the plate material is 3 to 10.

5. The edge detection method according to claim 3, wherein a width direction coordinate at which the differential intensity integration value obtained from the total length in the longitudinal direction of the captured image is maximized is used as a detection reference coordinate, and edge detection of each divided region is performed based on the detection reference coordinate.

6. The edge detection method according to claim 5, wherein a coordinate at which the differential integration value in each divided region is maximized within a range of ±5 to 40 mm of the detection reference coordinate is determined as an edge coordinate in each divided region.

7. Determining whether the differential intensity integration value is less than a threshold value, and if it is less than the threshold value, determining it as a value with low reliability, and using the differential intensity integration value of the image captured one before. The edge detection method according to any one of claims 1 to 6.

8. If the coordinates of the detected edge are different from the edge coordinates of the image captured one before by a value of a certain amount or more within a range of ±30 to 70 mm, it is determined as a value with low reliability, and the edge coordinates of the image captured one before are used. The edge detection method according to any one of claims 1 to 6.

9. An edge detection device for detecting an edge of a sheet-shaped plate material continuously imaged in the longitudinal direction of the plate material to be conveyed, comprising: a luminance differential value acquisition unit that acquires a luminance differential value in the width direction of the plate material from the captured image; A differential intensity integration value calculation unit that calculates a differential intensity integration value obtained by integrating the luminance differential value in the longitudinal direction of the plate material; An edge coordinate calculation unit that obtains the coordinates at which the differential intensity integration value is maximized; An edge detection device having the above, and based on the coordinates, the edge coordinates of the plate material are detected.

10. The edge detection device further includes an image division unit that divides the captured image into a plurality of parts in the longitudinal direction of the plate material, The luminance differential value acquisition unit acquires the luminance differential value in the width direction in each divided region divided by the division unit, the differential intensity integration value calculation unit obtains the differential intensity integration value of each divided region, and the edge coordinate calculation unit obtains the coordinates at which the partial intensity integration value in each divided region is maximized. The edge detection device according to claim 9.

11. The edge coordinate calculation unit determines, as the edge coordinates in each divided region, the coordinates at which the differential integration value in each divided region is maximized within a range of ±5 to 40 mm of the detection reference coordinates. The edge detection device according to claim 10.

12. The edge detection device according to any one of claims 9 to 11 further includes a differential intensity integration value determination unit that compares the differential intensity integration value with a preset threshold value and determines that the differential intensity integration value is an unreliable value when it is less than the threshold value.

13. The edge detection device according to any one of claims 9 to 11 further includes an edge coordinate determination unit that determines that the value is of low reliability if the coordinates of the detected edge are different by a value of a certain amount or more within a range of ±30 to 70 mm from the edge coordinates of the image captured one before, and uses the edge coordinates of the image captured one before.

14. A leveling control method for performing leveling control of a rolling device that rolls a sheet-like plate material to be conveyed, A step of detecting the edge coordinates of the plate material from an image continuously captured in the longitudinal direction of the conveyed sheet-like plate material; A step of calculating the amount of meandering and the amount of bending of the plate material based on the detected edge coordinates; A step of applying a leveling control signal to the rolling device based on the calculated amount of meandering and the amount of bending; Having The step of detecting the edge coordinates of the plate material is From the captured image, the luminance differential value in the width direction of the plate material is acquired. A leveling control method for obtaining coordinates at which an integrated differential intensity value obtained by integrating the luminance differential value in the longitudinal direction of the plate material is maximized, and detecting an edge coordinate of the plate material based on the coordinates.

Images