A method for quickly locating the edge of hot-rolled strip steel

Abstract

The invention provides a method for quickly locating the edge of a hot-rolled steel strip, comprising: using a camera to collect images of the steel strip between stands during the rolling process in real time; summing the grayscale values ​​of each column of pixel points in the steel strip image along the direction of movement of the steel strip to obtain a grayscale array; subtracting the previous item from the next item in the grayscale array to obtain a gradient array; finding the maximum value M among the first T1 elements in the gradient array; and searching backwards from the T1+1th element in the gradient array until b is found. j , so that from b j The values ​​of the three consecutive elements from the beginning are all greater than M; the jth column in the strip image is the edge position of one side of the strip; find the minimum value N among the last T3 elements in the gradient array; in the gradient array, start searching from the n-T3-1th element forward until b is found k , so that from b k The values ​​of the first three consecutive elements are all less than N; the kth column in the strip image is the edge position of the other side of the strip.

Description

Technical Field

[0001] The invention belongs to the technical field of hot-rolled strip position detection and relates to a method for quickly locating the edge of a hot-rolled strip. Background Art

[0002] Strip deviation is a significant factor affecting strip quality during the finishing process of hot strip milling. Strip deviation refers to the deviation of the strip centerline from the centerline of the rolling equipment during the finishing process due to lateral bending of the incoming material and intermediate bar, as well as precision issues with the rolling equipment. Strip deviation causes asymmetric operation of the rolling mill, manifesting as significant deviations in the rolling forces and roll gap width on the operating and drive sides. This interferes with the mill's shape control system and causes quality issues such as strip wavering. If strip deviation is not detected promptly, the deviation may worsen, leading to abnormal friction and collision between the strip and the rolls and guides. This not only affects the strip's surface quality but also damages production equipment. In severe cases, it can even cause "tail swinging" and "steel piling," impacting production schedule and safety.

[0003] Domestic and foreign researchers have conducted a lot of research on the detection of the edge position of the strip during the rolling process. The Chinese journal article "Research and Application of Asymmetric Measurement and Control System for Hot Rolling Operation" (Metallurgical Automation, 2020, 44(01): 48-54.) designed and developed an asymmetric detection and control system for hot rolling operation based on machine vision technology. For the deviation detection part of the finishing rolling, a binocular linear array camera is used for image acquisition to obtain the strip width and deviation information. The system has been successfully applied to the production practice of hot rolling. The application results show that the detection device operates stably, can perform online detection of finishing rolling deviation, and the detection accuracy meets the use requirements.

[0004] Chinese invention patent publication number CN113828641A, "A Machine Vision-Based Method for Processing the Deviation Curve Between Finishing Strip Stands," describes a method for processing the deviation curve between strip stands. This method uses a binocular linear array camera mounted on the top of the stand to capture the strip edge and calculate the strip width and deviation. The method then calculates the difference between the actual strip width measured by the exit width gauge and the measured strip width. Deviation data corresponding to width differences exceeding a threshold are filtered to remove noise from the deviation curve. Chinese invention patent publication number CN113465497A, "A Method for Online Detection of the Moving Position of Hot-Rolled Strip Overcoming Water Mist Interference," describes a method for online detection of the moving position of hot-rolled strip overcoming water mist interference. This method uses a grayscale linear upscaling algorithm to process images captured after water mist interference, resulting in clear and distinct strip edge images. This method avoids interference from water mist on strip edge detection and improves the reliability and effectiveness of the detection system.

[0005] The above-mentioned research institutes have two main deficiencies:

[0006] (1) The strip moves very fast during the rolling process, and the image data volume is very large, so real-time performance is required to be very high. Therefore, for edge detection operators such as Sobel and Roberts, which require the gradient of the entire image, methods such as the Canny edge detection algorithm that require statistics on local gradients of the image, methods using Hough line detection that require positional relationship analysis between pixels of the entire image, and methods using deep learning for edge extraction, the computational complexity is large, making it difficult to ensure the real-time and continuity of the detection results.

[0007] (2) The brightness of the environment in which the strip is located varies greatly, and there is interference on the surface of the strip. Therefore, the method of directly finding the maximum gradient value and the method of image binarization used in the algorithm will reduce the adaptability of the algorithm to the environment and affect the accuracy of the detection results. Summary of the Invention

[0008] In order to solve the above technical problems, the purpose of the present invention is to provide a method for quickly locating the edge of a hot-rolled strip, which takes into account the difference between the grayscale gradient change of the strip edge and the grayscale gradient change of the background noise, and can quickly locate the position of the strip edge with a smaller amount of calculation.

[0009] A method for quickly locating an edge of a hot-rolled strip steel according to the present invention comprises:

[0010] Step 1: Use a camera to capture real-time images of the strip steel between stands during the rolling process. The resolution of the strip steel image is n×m.

[0011] Step 2: Sum the grayscale values ​​of each column of pixels in the strip image along the strip movement direction to obtain the grayscale array {a1, a2, a3, ..., a n};

[0012] Step 3: Convert the grayscale array {a1, a2, a3, ..., a n}, we get the gradient array {b1,b2,b3,…,b n-1};

[0013] Step 4: Find the maximum value M among the first T1 elements in the gradient array. T1 is determined based on the position limit of the strip's lateral movement and the distribution of the grayscale sum.

[0014] Step 5: In the gradient array, search backward from the T1+1th element until b is found j , so that from b j The values ​​of the three consecutive elements after the beginning are all greater than M;

[0015] Step 6: The jth column in the strip image is the edge position of one side of the strip;

[0016] Step 7: Find the minimum value N among the reciprocal T3 elements in the gradient array. T3 is determined based on the position limit of the strip's lateral movement and the distribution of the grayscale sum.

[0017] Step 8: In the gradient array, search forward from the n-T3-1th element until b is found k , so that from b k The values ​​of the first three consecutive elements are all less than N;

[0018] Step 9: The kth column in the strip image is the edge position of the other side of the strip.

[0019] Furthermore, the strip image with a resolution of n×m collected in step 1 includes n pixels along the width direction of the strip and m pixels along the moving direction of the strip, and its pixel matrix is ​​as follows:

[0020]

[0021] Among them, p gh The gray value of the position corresponding to the gth row and hth column of the strip image is 1≤g≤m, 1≤h≤n; it is defined from p 1h to p mh The direction is vertical, that is, the direction of strip movement, from p g1 to p gn The direction is horizontal, that is, the width direction of the strip.

[0022] Furthermore, the grayscale array {a1, a2, a3, ..., a n Any element a in h Represented as a h =p 1h +p 2h +p gh …+p mh .

[0023] Furthermore, the gradient array {b1, b2, b3, ..., b n-1 Any element b in h Indicated as b h =a h+1 -a h .

[0024] Furthermore, in step 4, according to the position limit of the lateral movement of the strip steel, there is no strip steel in the area from the 1st column to the T1th column in the strip steel image, and T1 is set to 1000.

[0025] Furthermore, the step 5 of searching backward from the T1+1th element specifically includes searching in the direction of increasing subscripts of the gradient array elements.

[0026] Furthermore, in step 6: since the edge of the steel strip is vertical, the edge position of this side of the steel strip is located in the jth column in the steel strip image.

[0027] Furthermore, in step 7, according to the position limit of the lateral movement of the strip steel, there is no strip steel in the area from the n-T3 column to the nth column in the strip steel image, and T3 is set to 1500.

[0028] Furthermore, the step 8 of searching forward from the n-T3-1th element is specifically performed by searching in a direction in which the subscripts of the gradient array elements decrease.

[0029] Furthermore, in step 9: since the edge of the steel strip is vertical, the edge position of the other side of the steel strip is located in the kth column in the steel strip image.

[0030] The present invention provides a method for quickly locating the edge of a hot-rolled strip. By collecting images of the strip between racks, the method realizes quick positioning of the strip edge based on the grayscale relationship between the strip surface and the background and the positional relationship of the strip in the camera field of view, without using a complex edge extraction algorithm. The method ensures the real-time and stability of the detection algorithm, and is used to replace operating workers to realize real-time monitoring of the strip deviation during the rolling process, thereby ensuring the rhythm and safety of the strip production process and improving the quality of the strip products. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1This is a flow chart of a method for quickly locating the edge of a hot-rolled strip steel according to the present invention;

[0032] Figure 2 This is a schematic diagram of the camera installation position;

[0033] Figure 3 It is a schematic diagram of the strip edge and the direction of strip movement;

[0034] Figure 4 It is a schematic diagram of the grayscale array in the coordinate system;

[0035] Figure 5 This is a schematic diagram of the gradient array in the coordinate system. DETAILED DESCRIPTION

[0036] like Figure 1 As shown, a method for quickly locating the edge of a hot-rolled strip steel according to the present invention comprises:

[0037] Step 1: Use a camera to capture real-time images of the strip between stands during the rolling process;

[0038] In this embodiment, one installation position of the camera is as follows Figure 2 As shown: The camera is installed between two sets of frames, and the edge of the strip needs to be within the camera's field of view. The strip moves vertically in the camera's field of view.

[0039] The strip image with a resolution of n×m collected in step 1 includes n pixels along the width direction of the strip and m pixels along the moving direction of the strip. Its pixel matrix is ​​as follows:

[0040]

[0041] Among them, p gh The gray value of the position corresponding to the gth row and hth column of the strip image is 1≤g≤m, 1≤h≤n; it is defined from p 1h to p mh The direction is vertical, that is, the direction of strip movement, from p g1 to p gn The direction is the horizontal direction, that is, the width direction of the strip steel. In this embodiment, the resolution of the strip steel image captured by the camera is 8192×500.

[0042] The edge position of the strip and the direction of strip movement are shown as follows: Figure 3 The two thick lines represent the strip edges on the left and right sides of the image, respectively. During normal rolling, the strip moves vertically as indicated by the arrows. When the strip deviates, the strip edges on both sides move horizontally.

[0043] Step 2: Sum the grayscale values ​​of each column of pixels in the strip image along the strip movement direction to obtain the grayscale array {a1, a2, a3, ..., a n};

[0044] Grayscale array {a1,a2,a3,…,a n Any element a in h Represented as a h =p 1h +p 2h +p gh …+p mh .

[0045] In the specific implementation, the strip image with a resolution of 8192×500 collected in step 1 is summed along the moving direction of the strip, that is, the vertical direction defined in step 1, to obtain a grayscale array {a1, a2, a3, ..., a 8192}.

[0046] Set the grayscale of array {a1,a2,a3,…,a 8192}Draw in the coordinate system as Figure 4 As shown, the horizontal axis of the coordinate system corresponds to the subscript of the grayscale array, that is, the horizontal pixel; the vertical axis corresponds to the numerical value of the corresponding element in the grayscale array, that is, the pixel grayscale sum.

[0047] Step 3: Convert the grayscale array {a1, a2, a3, ..., a n}, we get the gradient array {b1,b2,b3,…,b n-1}; Gradient array {b1,b2,b3,…,b n-1 Any element b in h Indicated as b h =a h+1 -a h .

[0048] In the specific implementation, a gray gradient array {b1, b2, b3, ..., b 8191}. Draw it in the coordinate system as Figure 5 As shown, the horizontal axis of the coordinate system corresponds to the gradient array {b1,b2,b3,L,b n-1}; the vertical axis corresponds to the numerical value of the corresponding element in the gradient array, that is, the pixel grayscale gradient.

[0049] Step 4: Find the maximum value M among the first T1 elements in the gradient array. T1 is determined based on the position limit of the strip's lateral movement and the distribution of the grayscale sum.

[0050] In this embodiment, there is no strip in the area from the 1st column to the 1000th column in the strip image, and the limit position of the horizontal movement of the strip edge cannot reach this area. T1 is set to 1000, thereby obtaining M as 127.

[0051] Step 5: In the gradient array, start searching backward from the T1+1th element, i.e. element b T1+1 Start searching backward until b is found j , so that from b j The values ​​of the three consecutive elements after the beginning are all greater than M. That is, b j 、b j+1 ...b j+2 Both are greater than 127.

[0052] In a specific implementation, searching backward from the T1+1th element is specifically performed as follows: searching in the direction of increasing subscripts of the gradient array elements.

[0053] Step 6: The jth column in the strip image is the edge position of one side of the strip;

[0054] In a specific implementation, since the edge of the steel strip is vertical, the edge position of this side of the steel strip is located in the jth column in the steel strip image.

[0055] Step 7: Find the minimum value N among the reciprocal T3 elements in the gradient array. T3 is determined based on the position limit of the strip's lateral movement and the distribution of the grayscale sum.

[0056] According to the position limit of the lateral movement of the strip, there is no strip in the area from the n-T3 column to the nth column in the strip image, and T3 is set to 1500.

[0057] Step 8: In the gradient array, search forward from the n-T3-1th element until b is found k , so that from b k The values ​​of the first three consecutive elements are all less than N;

[0058] In a specific implementation, searching forward from the n-T3-1th element is specifically performed as follows: searching in a direction where the subscripts of the gradient array elements decrease.

[0059] Step 9: The kth column in the strip image is the edge position of the other side of the strip.

[0060] In a specific implementation, since the edge of the steel strip is vertical, the edge position of the other side of the steel strip is located in the kth column in the steel strip image.

[0061] like Figure 5It can be seen that the grayscale gradient change characteristics of the strip edge within a certain range are drastic and continuously of the same sign; while the grayscale gradient change characteristics of the background noise are gentle and cannot be continuously of the same sign. The method of the present invention utilizes the difference between the two characteristics to achieve the positioning of the strip edge: first, T1 (T3) is used to find the maximum (minimum) value of the gradient of the background noise within a certain range, that is, the gradient change threshold M (N) is adaptively extracted from the background noise; after finding the maximum (minimum) value of the gradient, since the grayscale gradient of the background changes gently and cannot be continuously of the same sign, it cannot meet the condition that T2 (T4) consecutive elements are continuously greater than (less than) the maximum (minimum) value of the gradient; however, the grayscale gradient of the strip edge changes drastically and continuously of the same sign, which can meet the condition that T2 (T4) consecutive elements are continuously greater than (less than) the maximum (minimum) value of the gradient.

[0062] In this algorithm, taking the edge positioning of the strip on one side as an example, two thresholds T1 and T2 need to be set. For T1, its significance is to define a certain background range so that the maximum value of the grayscale gradient can be searched within the background. As long as the value of T1 does not exceed the horizontal movement range of the strip, it will not affect the detection results. For T2, its significance is to distinguish the gradient changes of the background noise and the gradient changes of the strip edge. Since the maximum gradient value M has been searched within the T1 range, only a few grayscale gradients will exceed M in the entire background range, and the situation where T2 consecutive elements are greater than M can only occur at the edge of the slab. Therefore, as long as T2 is not set too large and exceeds the range where the grayscale gradient of the edge rises sharply, it can meet the detection requirements.

[0063] Combined with the positional relationship of the steel strip in the camera field of view, the method of the present invention can achieve rapid positioning of the edge of the steel strip without using a complex edge extraction algorithm, greatly reducing the amount of algorithm calculation, improving the algorithm's calculation speed, ensuring the algorithm's real-time performance and stability, and reducing the hardware requirements for image processing equipment.

[0064] The above description is only a preferred embodiment of the present invention and is not intended to limit the concept of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for quickly locating the edge of a hot-rolled strip, characterized in that: include: Step 1: Use a camera to capture real-time images of the strip steel between stands during the rolling process. The resolution of the strip steel image is n×m. Step 2: Sum the grayscale values ​​of each column of pixels in the strip image along the strip movement direction to obtain the grayscale array {a1, a2, a3, ..., a n }; Step 3: Convert the grayscale array {a1, a2, a3, ..., a n }, we get the gradient array {b1,b2,b3,…,b n-1 }; Step 4: Find the maximum value M among the first T1 elements in the gradient array. T1 is determined based on the position limit of the strip's lateral movement and the distribution of the grayscale sum. There is no strip in the area from column 1 to column T1 in the strip image. Step 5: In the gradient array, search backward from the T1+1th element until b is found j , so that from b j The values ​​of the three consecutive elements after the beginning are all greater than M; Step 6: The jth column in the strip image is the edge position of one side of the strip; Step 7: Find the minimum value N among the T3 elements in the gradient array. T3 is determined based on the position limit of the strip's lateral movement and the distribution of the grayscale sum. There is no strip in the area from the n-T3 column to the nth column in the strip image. Step 8: In the gradient array, search forward from the n-T3-1th element until b is found k , so that from b k The values ​​of the first three consecutive elements are all less than N; Step 9: The kth column in the strip image is the edge position of the other side of the strip.

2. The method for quickly locating the edge of a hot-rolled strip according to claim 1, wherein: The strip image with a resolution of n×m collected in step 1 includes n pixels along the width direction of the strip and m pixels along the moving direction of the strip. Its pixel matrix is ​​as follows: Among them, p gh The gray value of the position corresponding to the gth row and hth column of the strip image is 1≤g≤m, 1≤h≤n; it is defined from p 1h to p mh The direction is vertical, that is, the direction of strip movement, from p g1 to p gn The direction is horizontal, that is, the width direction of the strip.

3. The method for quickly locating the edge of a hot-rolled strip according to claim 2, wherein: In step 2, the grayscale array {a1, a2, a3, ..., a n Any element a in h Represented as a h =p 1h +p 2h +p gh …+p mh .

4. The method for quickly locating the edge of a hot-rolled strip according to claim 1, wherein: In step 3, the gradient array {b1, b2, b3, ..., b n-1 Any element b in h Indicated as b h =a h+1 -a h .

5. The method for quickly locating the edge of a hot-rolled strip according to claim 1, wherein: In step 4, according to the position limit of the lateral movement of the strip steel, there is no strip steel in the area from the 1st column to the T1th column in the strip steel image, and T1 is set to 1000.

6. The method for quickly locating the edge of a hot-rolled strip according to claim 1, wherein: The step 5 of searching backward from the T1+1th element specifically includes searching in the direction of increasing subscripts of the gradient array elements.

7. The method for quickly locating the edge of a hot-rolled strip according to claim 2, wherein: In step 6: since the edge of the steel strip is vertical, the edge position of this side of the steel strip is located in the jth column in the steel strip image.

8. The method for quickly locating the edge of a hot-rolled strip according to claim 1, wherein: In step 7, according to the position limit of the lateral movement of the strip, there is no strip in the area from the n-T3 column to the nth column in the strip image, and T3 is set to 1500.

9. The method for quickly locating the edge of a hot-rolled strip according to claim 1, wherein: The step 8 of searching forward from the n-T3-1th element is specifically: searching in the direction of decreasing subscripts of the gradient array elements.

10. The method for quickly locating the edge of a hot-rolled strip according to claim 2, wherein: In step 9: since the edge of the steel strip is vertical, the edge position of the other side of the steel strip is located in the kth column in the steel strip image.

Images