A method and system for automatically determining the length of cut-to-length steel plates based on machine vision

Abstract

The present invention discloses a method and system for automatic length determination of cut-to-length steel plates based on machine vision, the method comprising: installing a monitoring camera at the edge of a roller after cut-to-length; placing a ruler below the shooting position of the monitoring camera at the edge of the roller; obtaining the length of the steel plate to be cut to length, determining the camera responsible for measuring the current cut-to-length length by calculation, and sending a confirmation signal responsible for the cut-to-length measurement to the determined camera, while the other cameras do not work; when the camera receives the above confirmation signal, it calculates the pixel position in the camera where the cut-to-length steel plate should be when the cut-to-length length is reached, as a target point; then the steel plate area position in the real-time collected image is analyzed, and when the steel plate area position reaches the pixel position of the target point, the automatic length determination of the current cut-to-length steel plate is completed. The present invention requires low hardware cost, is easy to install and implement, and can achieve high detection accuracy. It is a solution with high cost performance to reduce labor costs and realize automatic cut-to-length cutting of steel plates.

Description

Technical Field

[0001] The invention relates to the technical field of machine vision detection, and in particular to an automatic length determination method and system for cut-to-length steel plates based on machine vision. Background Art

[0002] The post-processing process of the steel plate production line includes cutting heads, cutting tails, and sizing. When sizing steel plates, humans need to rely on monitoring cameras arranged on the rollers to observe the progress of the steel plates and set marks at commonly used fixed positions to achieve the purpose of judging the length of the steel plates. The human eye needs to observe the monitoring screen in real time to determine the position of the steel plate, which greatly limits the operation process. In addition, the subjectivity of manual operation can easily lead to practical problems such as the cutting accuracy not meeting the target when cutting steel plates to length.

[0003] Currently, the known methods for measuring the length of steel plates mainly rely on indirect measurement methods such as encoders and laser velocimeters. The encoder is generally installed on the roller shaft, and the length of the steel plate is indirectly calculated by tracking the circumference of the running roller. This method ignores the effects of steel plate sliding and poor contact, and the length measurement accuracy is low. The laser velocimeter directly measures the speed of the steel plate and has a high length measurement accuracy. It is a relatively reliable method for measuring the length of the steel plate. However, this type of equipment is generally expensive, with a price of more than 100,000 yuan, which is often not worth the cost for small needs. In summary, this method of obtaining the length based on speed measurement will introduce errors in time, and this error is often unpredictable and unavoidable. Summary of the invention

[0004] The present invention provides a method and system for automatically determining the length of a cut-to-length steel plate based on machine vision, which realizes the length measurement of the steel plate by using multiple groups of cameras distributed after the cut-to-length steel plate is cut to length, so as to solve the technical problems of low length measurement accuracy or expensive equipment and high cost in the prior art.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] On the one hand, the present invention provides a method for automatically determining the length of a cut-to-length steel plate based on machine vision, and the method for automatically determining the length of a cut-to-length steel plate based on machine vision comprises:

[0007] A plurality of monitoring cameras are installed at the edge of the roller after the cut-to-length cut; wherein the plurality of monitoring cameras are distributedly arranged behind the cut-to-length cut, and all distributed monitoring cameras are connected to the same local area network and are uniformly scheduled by a data server; the size of the area photographed by each monitoring camera in the width direction of the roller accounts for 0.3 to 0.5 of the entire roller width, and the coverage ranges of the monitoring cameras along the roller direction retain a certain overlap;

[0008] A scale is arranged below the shooting position of the monitoring camera at the edge of the roller path; wherein, the length of the scale is greater than the difference between the maximum length and the minimum length of the fixed-length steel plate, the scale is provided with identification points at a certain unit interval, and the identification points are marked with the length values from the scissors of the fixed-length shear.

[0009] Obtain the length of the steel plate to be determined by the data server, determine the camera responsible for the current fixed-length measurement through calculation, and send a confirmation signal for fixed-length measurement to the determined camera, and the remaining cameras do not work.

[0010] When the camera receives the confirmation signal for fixed-length measurement, adopt a real-time calibration strategy to obtain the resolution of the currently working camera, calculate the pixel position where the fixed-length steel plate should be in the camera when reaching the fixed length, and determine the target point based on the calculated pixel position; then collect the fixed-length steel plate image in real time, extract the position of the steel plate area in the collected image, and perform real-time analysis on the extracted position of the steel plate area. When the position of the steel plate area reaches the pixel position of the target point, the automatic length determination of the current fixed-length steel plate is completed.

[0011] Further, when installing multiple monitoring cameras at the edge of the roller path after the fixed-length shear, the number of cameras N is obtained by the following formula:

[0012]

[0013] wherein, L min is the minimum length of the fixed-length steel plate, L max is the maximum length of the fixed-length steel plate, l is the shooting range of the camera along the roller path direction, and μ is the overlap ratio between adjacent cameras.

[0014] The distribution position P of the cameras i is obtained by the following formula:

[0015] P i = L min +(i - 1)(1 - μ)l

[0016] wherein, i is the index number of the camera, and the index range is from 1 to N.

[0017] Further, the scale adopts a pure white background, the identification points adopt pure black rectangular marks, the position distance represented by the identification points is the actual distance from the identification points to the scissors of the fixed-length shear, and the interval distance d of each identification point < l / 2 to ensure that each monitoring camera can at least capture 2 identification points on the scale.

[0018] Further, the determination of the camera responsible for the current fixed-length measurement through calculation includes:

[0019] Calculate the index number n of the camera responsible for the current fixed-length measurement through the following formula:

[0020]

[0021] Wherein, L is the length of the steel plate to be sized, and [*] is the symbol for rounding down.

[0022] Furthermore, the real-time calibration strategy is used to obtain the resolution of the currently working camera, including:

[0023] When the camera receives the confirmation signal responsible for the fixed-length measurement, it immediately collects a real-time image. In the collected real-time image, the pixel position of the mark point in the image is extracted using the threshold segmentation method. The pixel position difference d between two adjacent mark points is calculated. pix , and then get the camera resolution σ:

[0024]

[0025] Where d is the distance between adjacent markers.

[0026] Furthermore, the pixel position of the cut-length steel plate in the camera that should be located when the cut-length steel plate reaches the cut-length length is calculated including:

[0027] The pixel position p where the cut-length steel plate will appear in the image is determined by the length L of the cut-length steel plate and the resolution of the camera responsible for measuring the current cut-length length:

[0028]

[0029] Among them, P′ j is the actual distance value represented by the point j in the camera responsible for the current fixed length measurement, p j is the pixel position value of the identification point j in the camera responsible for the current fixed length measurement.

[0030] Furthermore, when extracting the position of the steel plate area in the captured image, a preset steel plate cutout model is used to extract the position of the steel plate area in the image; the model is composed of an encoding part and a decoding part, the model input is a real-time captured steel plate image and a separate background image, the model output is a foreground probability map, a background probability map, a foreground error map and a foreground residual map, the grayscale value of the background area in the foreground probability map output by the model is 0, and the grayscale value of the steel plate area is 1, and the starting point of its horizontal projection is the real-time pixel position of the steel plate head.

[0031] Further, determining the target point based on the calculated pixel position includes:

[0032] A preset compensation amount is added based on the calculated pixel position where the cut-length steel plate should be located in the camera when reaching the cut-length, so as to reduce the influence of the steel plate movement inertia and time delay on the cut-length.

[0033] Furthermore, after the automatic length determination of the current cut-length steel plate is completed, the method further comprises:

[0034] The signal of steel plate length determination is sent to the first-level automation system, and the steel plate cutting operation is performed by the cut-to-length shear.

[0035] On the other hand, the present invention also provides a machine vision-based automatic length determination system for steel plates cut to length, the automatic length determination system for steel plates cut to length comprising: a data server, a plurality of monitoring cameras and a ruler; wherein,

[0036] The multiple monitoring cameras are installed at the edge of the roller after the cut-to-length shear; and the multiple monitoring cameras are distributed and arranged behind the cut-to-length shear, and all the distributed monitoring cameras are connected to the same local area network and are uniformly scheduled by the data server; the size of the area photographed by each monitoring camera in the width direction of the roller accounts for 0.3 to 0.5 of the entire roller width, and the coverage range of each monitoring camera along the roller direction retains a certain overlap;

[0037] The ruler is placed below the shooting position of the monitoring camera at the edge of the roller; wherein the length of the ruler is greater than the difference between the maximum length and the minimum length of the cut-to-length steel plate, and marking points are arranged on the ruler at intervals of a certain unit, and the marking points are marked with the length values ​​of the distances from the cut-to-length shears;

[0038] The data server is used to perform the following steps:

[0039] Obtain the length of the steel plate to be cut, determine the camera responsible for measuring the current cut length through calculation, and send a confirmation signal to the determined camera responsible for the cut length measurement, while the other cameras do not work;

[0040] A real-time calibration strategy is used to obtain the resolution of the currently working camera, and the pixel position of the cut-length steel plate in the camera should be calculated when the cut-length steel plate reaches the cut-length, and the target point is determined based on the calculated pixel position;

[0041] The currently working camera is controlled to collect the image of the cut-length steel plate in real time, and the steel plate area position in the collected image is extracted, and the extracted steel plate area position is analyzed in real time. When the steel plate area position reaches the pixel position of the target point, the automatic length determination of the current cut-length steel plate is completed.

[0042] The beneficial effects brought about by the technical solution provided by the present invention include at least:

[0043] The automatic length determination method and system for cut-to-length steel plates based on machine vision proposed in the present invention provide a new solution for automatic length determination after cut-to-length. The technical solution of the present invention utilizes a monitoring camera and a ruler to simulate the human eye to judge the length of the steel plate, and directly measures the length to avoid the time error problem introduced by traditional encoders and laser speed meters during measurement. In addition, the hardware cost required for the solution of the present invention is relatively low, and it is more suitable for cut-to-length measurement in general scenarios, and the accuracy can be easily met by designing the camera selection. The solution of the present invention can replace manual labor, achieve the purpose of automatic cut-to-length, and increase the production efficiency of the production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0045] Figure 1 It is a schematic diagram of the execution flow of the automatic length determination method for cutting steel plates to length based on machine vision provided by an embodiment of the present invention;

[0046] Figure 2 is a schematic diagram of a steel plate length-fixing process provided by an embodiment of the present invention;

[0047] Figure 3 is a distribution diagram of camera installation positions provided by an embodiment of the present invention;

[0048] Figure 4 is a schematic diagram of the positions of a camera and a ruler provided in an embodiment of the present invention;

[0049] Figure 5 It is a schematic diagram of the steel plate cutout algorithm flow provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0051] First embodiment

[0052] This embodiment provides a method for automatic length determination of cut-to-length steel plates based on machine vision, which realizes the length measurement of steel plates through multiple groups of cameras distributed behind the cut-to-length shears. The hardware equipment used in this method consists of a ruler with distance marking points installed at the edge of the roller, and monitoring cameras distributed and arranged behind the cut-to-length shears. The length of the steel plate to be cut is obtained through the automated secondary system, and based on this, it is determined which specific camera is responsible for the cut-to-length measurement of the steel plate; the camera responsible for the measurement first performs self-calibration, and then calculates the target pixel position that the cut-to-length steel plate should reach; then the camera performs real-time image detection, and uses a self-designed steel plate cutout model to output the steel plate foreground area; when the head projection point of the steel plate area reaches the target pixel position, the cut-to-length shear is triggered to cut the steel plate, thereby achieving the purpose of automatic cut-to-length cutting of the steel plate.

[0053] Specifically, the execution process of this method is as follows Figure 1 As shown, the following steps are included:

[0054] S1, multiple monitoring cameras are installed at the edge of the roller after the cut-to-length shear; wherein the multiple monitoring cameras are distributed and arranged behind the cut-to-length shear, and all distributed monitoring cameras are connected to the same local area network and are uniformly scheduled by the data server; the size of the area photographed by each monitoring camera in the width direction of the roller accounts for 0.3 to 0.5 of the entire roller width, and the coverage range of each monitoring camera along the roller direction retains a certain overlap;

[0055] Specifically, in this embodiment, Figure 2 As shown in the figure, taking the running roller of 3500mm as an example, a 6-megapixel monitoring camera is installed at the edge of the roller after the cut-to-length, and the area captured by the camera in the width direction of the roller is 1500mm. Figure 3 As shown in the figure, the coverage range between the cameras along the roller direction retains a certain overlap, and the overlap is 5% of the camera shooting width; the distribution position and number of cameras are determined according to the minimum length and maximum length of the cut-length steel plate, where the number of cameras N is obtained by the following formula:

[0056]

[0057] Among them, L min is the minimum length of the cut-to-size steel plate, L max is the maximum length of the cut-to-size steel plate, l is the shooting range of the camera along the roller direction, and μ is the overlap ratio between adjacent cameras, such as Figure 4 shown.

[0058] Camera distribution position P i (i∈[1,N]) is obtained by the following formula:

[0059] P i =L min+(i - 1)(1 - u)l

[0060] Among them, i is the index number of the camera, and the index range is from 1 to N.

[0061] S2. Place a scale below the shooting position of the monitoring camera at the edge of the roller table; among them, the length of the scale is greater than the difference between the maximum length and the minimum length of the fixed-length steel plate, the scale is provided with identification points at a certain unit interval, and the identification points are marked with the length value from the shears of the fixed-length shear.

[0062] Specifically, in this embodiment, the scale uses a pure white background, the identification points use pure black rectangular marks, the position distance represented by the identification points is the actual distance from the identification points to the shears of the fixed-length shear, and the interval distance d of each identification point is < l / 2 to ensure that each monitoring camera can at least capture 2 identification points on the scale.

[0063] S3. Obtain the length of the steel plate to be fixed-length from the data server, determine the camera responsible for the current fixed-length measurement through calculation, and send a confirmation signal for fixed-length measurement to the determined camera, and the other cameras do not work.

[0064] It should be noted that, in this embodiment, the length of the steel plate to be fixed-length is received by the data server from the automation secondary system; the method for determining the camera responsible for the current fixed-length measurement through calculation is as follows:

[0065] Calculate the index number n of the camera responsible for the current fixed-length measurement through the following formula:

[0066]

[0067] Among them, L is the length of the steel plate to be fixed-length, and [*] is the symbol for rounding down.

[0068] S4. When the camera receives the confirmation signal for fixed-length measurement, adopt a real-time calibration strategy to obtain the resolution of the currently working camera, calculate the pixel position where the fixed-length steel plate should be in the camera when reaching the fixed-length, and determine the target point based on the calculated pixel position; then collect the fixed-length steel plate image in real time, extract the position of the steel plate area in the collected image, and perform real-time analysis on the extracted position of the steel plate area. When the position of the steel plate area reaches the pixel position of the target point, the automatic length determination of the current fixed-length steel plate is completed.

[0069] Specifically, in this embodiment, the method for obtaining the resolution of the currently working camera by adopting a real-time calibration strategy is: when the camera receives the confirmation signal for fixed-length measurement, immediately collect a real-time image, use the threshold segmentation method in the collected real-time image to extract the pixel positions of the identification points in the image, and randomly select two adjacent position identification points to calculate their pixel position difference d pix, and then get the camera resolution σ:

[0070]

[0071] Where d is the distance between adjacent markers.

[0072] The method for calculating the pixel position of the cut-length steel plate in the camera when the cut-length steel plate reaches the cut-length length is as follows:

[0073] The pixel position p where the cut-length steel plate will appear in the image is determined by the length L of the cut-length steel plate and the resolution of the camera responsible for measuring the current cut-length length:

[0074]

[0075] Among them, P′ j is the actual distance value represented by the point j in the camera responsible for the current fixed length measurement, p j is the pixel position value of the identification point j in the camera responsible for the current fixed length measurement.

[0076] It should also be noted that after calculating the pixel position in the camera where the cut-length steel plate should be when the cut-length length is reached, theoretically the calculated pixel position can be directly used as the target point; however, in actual production, there are steel plate movement inertia and system time delay. Therefore, in order to reduce the impact of steel plate movement inertia and time delay on the cut-length, a certain compensation amount will be added when defining the target point pixel position of the steel plate in camera n (the camera responsible for the current cut-length measurement). The compensation amount is determined based on actual conditions.

[0077] In addition, when extracting the steel plate area position in the collected image, this embodiment uses a self-designed steel plate cutout model to extract the steel plate area position in the collected image; the steel plate cutout model is as follows: Figure 5 As shown in FIG. 1 , the model is composed of an encoding part and a decoding part. The model input is a steel plate image and a separate background image acquired in real time. The model output is a foreground probability map, a background probability map, a foreground error map and a foreground residual map.

[0078] The definition of the acquired image is composed of the following formula:

[0079] I=αF+(1-α)B

[0080] Among them, I is the collected image, B is the pure background image, F is the pure foreground image, and α is the foreground probability image.

[0081] The design model structure is shown in Table 1:

[0082] Table 1 Model structure

[0083]

[0084]

[0085] The optimization loss function L of the entire model is defined as:

[0086] L=L α +L β +L F +L E

[0087] Among them, L α is the foreground probability map loss function:

[0088]

[0089] in, is the foreground probability image predicted by the model, α is the target foreground probability image, To obtain the Sobel gradient.

[0090] L β is the background probability map loss function:

[0091]

[0092] β=1-α

[0093] in, is the background probability image predicted by the model, β is the target background probability image, To obtain the Sobel gradient.

[0094] L F Foreground image loss function:

[0095]

[0096] in, is the foreground image predicted by the model, and F is the target foreground image.

[0097] L E is the foreground probability error map loss function:

[0098]

[0099] in, is the foreground error image predicted by the model, and E is the target foreground error image.

[0100] In the foreground probability map output by the model, the grayscale value of the background area is 0, and the grayscale value of the steel plate area is 1. The starting point of its horizontal projection is the real-time pixel position of the steel plate head.

[0101] Furthermore, after the automatic length determination of the current cut-length steel plate is completed, the method further comprises:

[0102] S5, sends a signal to the first-level automation system that the steel plate has been cut to length, and the cut-to-length shear performs the cutting operation.

[0103] In summary, the method of this embodiment directly starts from the length measurement, avoiding the problem of being unable to estimate the time error when calculating the length due to the speed accumulation. A machine vision-based method is used to replace the human eye observation to determine the position of the steel plate to determine the fixed length position of the steel plate. By distributing and arranging cameras after the cut-to-length cutting, the actual position of the steel plate on the roller is obtained, and the plate length is calculated by comparing it with the ruler position. This method has low hardware cost and directly measures the length. The measurement accuracy is higher than that of the method through speed measurement. It is very effective for the fixed length of the steel plate at the cut-to-length cutting, and can achieve the purpose of automatic cut-to-length cutting.

[0104] Second embodiment

[0105] This embodiment provides a machine vision-based automatic length determination system for steel plates cut to length, which includes: a data server, multiple monitoring cameras and a ruler; wherein:

[0106] The multiple monitoring cameras are installed at the edge of the roller after the cut-to-length shear; and the multiple monitoring cameras are distributed and arranged behind the cut-to-length shear, and all the distributed monitoring cameras are connected to the same local area network and are uniformly scheduled by the data server; the size of the area photographed by each monitoring camera in the width direction of the roller accounts for 0.3 to 0.5 of the entire roller width, and the coverage range of each monitoring camera along the roller direction retains a certain overlap;

[0107] The ruler is placed below the shooting position of the monitoring camera at the edge of the roller; wherein the length of the ruler is greater than the difference between the maximum length and the minimum length of the cut-to-length steel plate, and marking points are arranged on the ruler at intervals of a certain unit, and the marking points are marked with the length values ​​of the distances from the cut-to-length shears;

[0108] The data server is used to perform the following steps:

[0109] Obtain the length of the steel plate to be cut, determine the camera responsible for measuring the current cut length through calculation, and send a confirmation signal to the determined camera responsible for the cut length measurement, while the other cameras do not work;

[0110] A real-time calibration strategy is used to obtain the resolution of the currently working camera, and the pixel position of the cut-length steel plate in the camera should be calculated when the cut-length steel plate reaches the cut-length, and the target point is determined based on the calculated pixel position;

[0111] The currently working camera is controlled to collect the image of the cut-length steel plate in real time, and the steel plate area position in the collected image is extracted, and the extracted steel plate area position is analyzed in real time. When the steel plate area position reaches the pixel position of the target point, the automatic length determination of the current cut-length steel plate is completed.

[0112] The automatic length-fixing system for cut-to-length steel plates based on machine vision of this embodiment corresponds to the automatic length-fixing method for cut-to-length steel plates based on machine vision of the first embodiment; wherein, the functions implemented by each functional module in the automatic length-fixing system for cut-to-length steel plates based on machine vision of this embodiment correspond one-to-one to each process step in the above-mentioned automatic length-fixing method for cut-to-length steel plates based on machine vision; therefore, they will not be repeated here.

[0113] In addition, it should be noted that the present invention can be provided as a method, an apparatus or a computer program product. Therefore, the embodiments of the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0114] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0115] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0116] It should also be noted that, in this article, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0117] Finally, it should be noted that the above is a preferred embodiment of the present invention. It should be pointed out that although the preferred embodiment of the present invention has been described, for those skilled in the art, once the basic creative concept of the present invention is known, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention. Therefore, the attached claims are intended to be interpreted as including the preferred embodiment and all changes and modifications that fall within the scope of the embodiments of the present invention.

Claims

1. A method for automatically determining the length of cut-to-length steel plates based on machine vision, characterized in that: Including: Installing a plurality of monitoring cameras at the edge of the roller table behind the fixed-length shears; wherein, the plurality of monitoring cameras are distributed behind the fixed-length shears, and all the distributed monitoring cameras are connected to the same local area network and are uniformly scheduled by a data server; the area size captured by each monitoring camera in the width direction of the roller table accounts for 0.3 to 0.5 of the entire width of the roller table, and there is a certain overlap in the coverage range along the roller direction between each monitoring camera; Placing a scale under the shooting position of the monitoring camera at the edge of the roller table; wherein, the length of the scale is greater than the difference between the maximum length and the minimum length of the fixed-length steel plate, and the scale is provided with identification points at a certain unit interval, and the identification points are marked with the length value from the shears of the fixed-length shears; The data server obtains the length of the steel plate to be cut to length, determines the camera responsible for the current fixed-length measurement through calculation, and sends a confirmation signal for fixed-length measurement to the determined camera, and the other cameras do not work; When the camera receives the confirmation signal for fixed-length measurement, it adopts a real-time calibration strategy to obtain the resolution of the currently working camera, calculates the pixel position where the fixed-length steel plate should be in the camera when the fixed-length is reached, and determines the target point based on the calculated pixel position; then it real-time collects the image of the fixed-length steel plate, extracts the position of the steel plate area in the collected image, and performs real-time analysis on the extracted position of the steel plate area. When the position of the steel plate area reaches the pixel position of the target point, the automatic fixed-length of the current fixed-length steel plate is completed; When installing a plurality of monitoring cameras at the edge of the roller table behind the fixed-length shears, the number of cameras N is obtained by the following formula: Among them, L min is the minimum length of the cut-to-size steel plate, L max is the maximum length of the cut-to-size steel plate, l is the shooting range of the camera along the roller direction, and μ is the overlap ratio between adjacent cameras; Camera distribution position P i Obtained by the following formula: P i =L min +(i-1)(1-μ)l Wherein, i is the index number of the camera, and the index range is from 1 to N; The calculating the pixel position where the fixed-length steel plate should be in the camera when the fixed-length is reached includes: Determining the pixel position p where the fixed-length steel plate will appear in the image through the length L of the fixed-length steel plate and the resolution of the camera responsible for the current fixed-length measurement: Among them, P′ j is the actual distance value represented by the point j in the camera responsible for the current fixed length measurement, p j is the pixel position value of the identification point j in the camera responsible for the current fixed length measurement.

2. The automatic length determination method for cut-to-length steel plates based on machine vision according to claim 1, characterized in that: The scale uses a pure white background, the identification points use pure black rectangular marks, the position distance represented by the identification points is the actual distance from the identification points to the shears of the fixed-length shears, and the interval distance d between each identification point < l / 2 to ensure that each monitoring camera can at least capture 2 identification points on the scale.

3. The automatic length determination method for cut-to-length steel plates based on machine vision according to claim 1, characterized in that: The determining the camera responsible for the current fixed-length measurement through calculation includes: Calculating the index number n of the camera responsible for the current fixed-length measurement through the following formula: Wherein, L is the length of the steel plate to be cut to length, and [*] is the symbol of rounding down.

4. The automatic length determination method for cut-to-length steel plates based on machine vision according to claim 1, characterized in that: The adopting the real-time calibration strategy to obtain the resolution of the currently working camera includes: When the camera receives the confirmation signal responsible for the fixed-length measurement, it immediately collects a real-time image. In the collected real-time image, the pixel position of the mark point in the image is extracted using the threshold segmentation method. The pixel position difference d between two adjacent mark points is calculated. pix , and then get the camera resolution σ: Wherein, d is the interval distance between adjacent identifications.

5. The automatic length determination method for cut-to-length steel plates based on machine vision according to claim 1, characterized in that: When extracting the position of the steel plate area in the collected image, a preset steel plate matte model is used to extract the position of the steel plate area in the image; The steel plate matte model consists of an encoding part and a decoding part. The model input is the steel plate image collected in real time and a separate background image. The model output is a foreground probability map, a background probability map, a foreground error map, and a foreground residual map. The gray value of the background area in the foreground probability map output by the model is 0, and the gray value of the steel plate area is 1. The starting point of its projection in the horizontal direction is the real-time pixel position of the head of the steel plate.

6. The automatic length determination method for cut-to-length steel plates based on machine vision according to claim 1, characterized in that: The step of determining the target point based on the calculated pixel position comprises: A preset compensation amount is added based on the calculated pixel position where the cut-length steel plate should be located in the camera when reaching the cut-length, so as to reduce the influence of the steel plate movement inertia and time delay on the cut-length.

7. The automatic length determination method for cut-to-length steel plates based on machine vision according to any one of claims 1 to 6, characterized in that: After the automatic length determination of the current cut-length steel plate is completed, the method further comprises: The signal of steel plate length determination is sent to the first-level automation system, and the steel plate cutting operation is performed by the cut-to-length shear.

8. An automatic length-fixing system for steel plates cut to length based on machine vision, characterized in that: The automatic length-fixing system for cutting steel plates to length comprises: a data server, a plurality of monitoring cameras and a ruler; wherein, The multiple monitoring cameras are installed at the edge of the roller after the cut-to-length shear; and the multiple monitoring cameras are distributed and arranged behind the cut-to-length shear, and all the distributed monitoring cameras are connected to the same local area network and are uniformly scheduled by the data server; the size of the area photographed by each monitoring camera in the width direction of the roller accounts for 0.3 to 0.5 of the entire roller width, and the coverage range of each monitoring camera along the roller direction retains a certain overlap; The ruler is placed below the shooting position of the monitoring camera at the edge of the roller; wherein the length of the ruler is greater than the difference between the maximum length and the minimum length of the cut-to-length steel plate, and marking points are arranged on the ruler at intervals of a certain unit, and the marking points are marked with the length values ​​of the distances from the cut-to-length shears; The data server is used to perform the following steps: Obtain the length of the steel plate to be cut, determine the camera responsible for measuring the current cut length through calculation, and send a confirmation signal responsible for the cut length measurement to the determined camera, while the other cameras do not work; A real-time calibration strategy is used to obtain the resolution of the currently working camera, and the pixel position of the cut-length steel plate in the camera should be calculated when the cut-length steel plate reaches the cut-length, and the target point is determined based on the calculated pixel position; Control the currently working camera to collect the image of the cut-length steel plate in real time, extract the steel plate area position in the collected image, and perform real-time analysis on the extracted steel plate area position. When the steel plate area position reaches the pixel position of the target point, the automatic length determination of the current cut-length steel plate is completed; When multiple monitoring cameras are installed at the edge of the roller after the cut-to-length process, the number of cameras N is obtained by the following formula: Among them, L min is the minimum length of the cut-to-size steel plate, L max is the maximum length of the cut-to-size steel plate, l is the shooting range of the camera along the roller direction, and μ is the overlap ratio between adjacent cameras; Camera distribution position P i Obtained by the following formula: P i =L min +(i-1)(1-μ)l Where i is the index number of the camera, and the index range is 1 to N; The calculation of the pixel position in the camera where the cut-length steel plate should be located when the cut-length steel plate reaches the cut-length includes: The pixel position p where the cut-length steel plate will appear in the image is determined by the length L of the cut-length steel plate and the resolution of the camera responsible for measuring the current cut-length length: Among them, P′ j is the actual distance value represented by the point j in the camera responsible for the current fixed length measurement, p j is the pixel position value of the identification point j in the camera responsible for the current fixed length measurement.

Images