Bimetal steel strip linear calibration detection system based on three-camera grouping measurement

Abstract

The invention relates to a bimetal steel strip linear calibration detection system based on three-camera grouping measurement, and relates to the field of metal processing calibration, and the bimetal steel strip linear calibration detection system comprises a camera measurement module which is provided with a first camera, a second camera and a third camera which are arranged at equal intervals and are used for collecting edge images of a bimetal steel strip at three different positions; the precision control module is configured with a camera precision matching strategy and is used for adjusting the detection precision of the edge images acquired by the first camera, the second camera and the third camera; the image optimization module is configured with a lens distortion calibration strategy and is used for performing distortion correction on an image acquired by the camera; and the image processing module is configured with a deviation adjusting strategy and used for calculating and analyzing the straightness deviation value of the edge image so as to determine the shearing deviation of the bimetallic steel strip and carrying out feedback adjustment. The double-metal steel strip shearing size monitoring device has the effects that the shearing size monitoring accuracy of the double-metal steel strip is improved, so that the time consumption of secondary machining is reduced, and the working efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of metal processing calibration technology, and in particular to a bimetallic steel strip linear calibration detection system based on three-camera group measurement. Background Art

[0002] Bimetallic steel strip is a composite metal steel strip with excellent physical properties such as high strength, impact resistance and corrosion resistance. In modern industrial production, bimetallic steel strip is formed into the required size through shearing.

[0003] In related technologies, during the cutting process of bimetallic steel strips of different sizes, in order to ensure that the cut size is within the set standard size error range, the bimetallic steel strips are directly detected and calibrated through a mechanical device combined with an optical fiber sensor. However, when faced with the detection and calibration work of different processing sizes, it is difficult for the optical fiber sensor to adjust the detection and calibration accuracy, and there is a large detection error, which requires secondary processing.

[0004] Regarding the above-mentioned related technologies, the accuracy of optical fiber sensors when calibrated with mechanical devices is greatly limited, and they cannot adapt to and adjust the accuracy range. As a result, bimetallic steel strips of different sizes are prone to errors exceeding standard requirements during shearing, which is not conducive to the accurate processing of products. The secondary shearing process prolongs the working time and the overall work efficiency is low. Summary of the Invention

[0005] In order to improve the accuracy of monitoring the shear size of bimetallic steel strips, reduce the time consumption of secondary processing, and improve work efficiency, the present application provides a bimetallic steel strip linear calibration detection system based on three-camera group measurement.

[0006] In a first aspect, the present application provides a bimetallic steel strip linear calibration and detection system based on three-camera group measurement, which adopts the following technical solutions: A bimetallic steel strip linear calibration detection system based on three-camera group measurement, comprising: A camera measurement module is provided with a first camera, a second camera, and a third camera arranged at equal intervals along the running path of the bimetallic steel strip after shearing, for respectively capturing edge images of the bimetallic steel strip at three different positions; an accuracy control module connected to the camera measurement module and configured with a camera accuracy matching strategy for adjusting the detection accuracy of edge images captured by the first camera, the second camera, and the third camera; An image optimization module, connected to the camera measurement module, and configured with a lens distortion calibration strategy for performing distortion correction on the camera-captured image; The image processing module is connected to the camera measurement module and is configured with a deviation adjustment strategy for calculating and analyzing the straightness deviation value of the edge image to determine the shear deviation of the bimetallic steel strip and perform feedback adjustment.

[0007] By adopting the above-mentioned technical solution, multiple modules cooperate with each other to complete the collaborative work of modules such as camera measurement precision control, lens distortion calibration and deviation adjustment. Combined with edge image acquisition and processing, it helps to improve the detection accuracy and calibration reliability of the straightness of bimetallic steel strips, adapt to the detection requirements of steel strips of different specifications, improve the stability of edge feature recognition, avoid the accuracy limitations of traditional mechanical calibration, reduce secondary processing caused by calibration errors, and improve the first-time pass rate and overall efficiency of steel strip shearing.

[0008] Optionally, the deviation adjustment strategy includes: Performing feature recognition based on the edge image to determine position values ​​corresponding to edge feature positions of the bimetallic steel strip in the first, second, and third camera fields of view, the position values ​​including a first position value, a second position value, and a third position value; The straightness deviation value of the bimetallic steel strip is calculated based on the position value and the preset deviation calculation model, and compared with the preset standard straightness deviation range to determine the deviation result; The shearing position of the bimetallic steel strip is dynamically feedback-adjusted according to the deviation result to keep the straightness deviation value within the preset standard straightness deviation range.

[0009] By adopting the above technical solution, the deviation adjustment strategy determines the edge position value through feature recognition, calculates the deviation based on the model and dynamically adjusts it, which helps to accurately capture the edge position of the steel strip, realize quantitative detection and real-time feedback of the straightness deviation, ensure that the deviation value is within the standard range, avoid shearing size errors caused by excessive deviation, and improve the dimensional consistency of steel strip shearing.

[0010] Optionally, the deviation calculation model adopts the following formula: ; ; ; ; in, Indicates the straightness deviation value of the bimetallic steel strip, 、 、 Represent the first position value, the second position value and the third position value respectively, is the position compensation coefficient, is the distortion correction coefficient, is the light source influence coefficient, is the lateral deviation between the actual position of the second camera and the midpoint of the line connecting the first and third cameras, is the preset reference deviation value, is the preset position influence coefficient, is the difference between the actual deviation value of the edge pixel of the image and the theoretical distortion-free value after lens distortion calibration. is the preset reference distortion pixel value, is the distortion influence coefficient, Grayscale uniformity of the edge area of ​​the bimetallic steel strip under parallel light source illumination, is the preset lower limit of the effective uniformity of the light source, is the upper limit of the ideal uniformity of the light source, It is the preset light source influence coefficient.

[0011] By adopting the above technical solution, the deviation calculation model introduces the position compensation coefficient K1, the distortion correction coefficient K2, and the light source influence coefficient K3. Combined with specific formulas, it quantitatively compensates for various errors. This helps to comprehensively eliminate the influence of installation position deviation, lens distortion, and uneven light source on measurement, improve the calculation accuracy of the straightness deviation value D, and ensure that the deviation detection results are more in line with reality.

[0012] Optionally, the camera accuracy matching strategy includes: Analyzing the straightness deviation value to determine a deviation range, and matching a corresponding lens model in a preset accuracy database according to the deviation range, wherein the lens model has different shooting field angles; The optimal shooting lens of the bimetallic steel strip of the current size specification is determined according to the corresponding lens models of the first camera, the second camera and the third camera, and the shooting lens corresponding to the lens model is switched based on the optimal shooting lens.

[0013] By adopting the above technical solution, the camera precision matching strategy matches the lens model according to the deviation range and switches to the optimal lens, which helps to achieve dynamic adaptation of the lens to the steel strip size and deviation requirements, avoids the problem of insufficient accuracy of a single lens in the detection of multiple specifications of steel strips, and improves the adaptability and precision stability of the detection of steel strips of different sizes.

[0014] Optionally, the camera measurement module is further configured with a camera position pre-calibration strategy, including: A standard coordinate axis orientation is established according to the moving direction of the bimetallic steel strip, and coordinate analysis is performed on the first camera and the third camera according to the standard coordinate axis orientation, and a reference line connecting the coordinates of the first camera and the second camera is generated; Perform point-line distance analysis based on the corresponding coordinates of the baseline and the second camera located in the middle to determine the error distance between the second camera and the baseline; A position compensation coefficient is generated based on the error distance and a preset position compensation model, and the second position value detected by the second camera is corrected according to the position compensation coefficient.

[0015] By adopting the above technical solution, the camera position pre-calibration strategy generates a position compensation coefficient through coordinate analysis and error distance calculation, which helps to eliminate the position error of camera installation in advance, reduce the interference of the second camera position deviation on the measurement, improve the baseline consistency of position value detection, and provide more reliable raw data for subsequent deviation calculation.

[0016] Optionally, the lens distortion calibration strategy includes: performing sub-pixel corner detection based on the captured edge image to determine lens distortion parameters of the first camera, the second camera, and the third camera; A polynomial distortion model is constructed based on the distortion parameters, reverse coordinate mapping is performed on the edge image of the bimetallic steel strip acquired in real time, and a pixel-level distortion compensation value is calculated; The camera field of view angle and edge feature position are dynamically corrected according to the distortion compensation value, so as to keep the compensated straight line measurement error less than the preset upper limit deviation value, and the corrected image is smoothed to maintain the edge feature clarity.

[0017] By adopting the above technical solution, the lens distortion calibration strategy helps to accurately eliminate edge pixel deviations caused by lens distortion through sub-pixel corner detection, polynomial distortion model correction and image smoothing processing, improve the accuracy of edge feature position recognition, ensure that straight line measurement errors are controlled within the preset range, and provide high-quality image data for deviation calculation.

[0018] Optionally, a lens field of view enhancement sub-strategy is also included, which uses the following steps to optimize and adjust the shooting field of view angle of the camera measurement module: Matching the corresponding optimal lens field of view angle in the preset database according to the shearing parameters of the bimetallic steel strip, and adjusting the shooting field of view angles of the first camera, the second camera, and the third camera; Monitor the changes in shearing parameters to determine the triggering of the field of view angle change instruction, and update the optimal field of view lens angle according to the changed shearing parameters. Perform image analysis based on the updated optimal field of view lens angle with a preset verification sub-strategy to keep the camera's image capture size within a preset standard size range.

[0019] By adopting the above technical solution, the lens field of view enhancement sub-strategy optimizes the field of view angle according to the shearing parameters and dynamically updates it, which helps to achieve real-time adaptation of the field of view angle to the steel strip shearing specifications, avoid edge omissions or insufficient accuracy caused by an excessively large or small field of view, maintain the rationality of the image shooting size, and improve the adaptability of multi-specification steel strip detection.

[0020] Optionally, the verification sub-strategy includes: Determining a width pixel range, a height pixel range, and an edge feature clarity threshold of an optimal image based on preset standard image size parameters, wherein the standard image size parameters correspond to shearing parameters of the bimetallic steel strip; Collect the bimetallic steel strip image at the current field of view angle, and extract the actual width pixel value, actual height pixel value and edge feature clarity value of the image; Based on the comparison between the actual size value and the standard size parameters, the size deviation value and clarity deviation value are calculated. If both are within the preset allowable deviation range, it is determined that the current field of view angle meets the requirements; if they exceed, the camera field of view angle is fine-tuned according to the deviation value and the image is recaptured until the image size and clarity meet the standard parameter requirements.

[0021] By adopting the above technical solution, the verification sub-strategy helps ensure that the image at the current field of view angle meets the detection requirements by comparing the size and clarity parameters of the actual and standard images and dynamically fine-tuning them, avoiding measurement errors caused by field of view deviation, providing a qualified image basis for subsequent edge recognition and deviation calculation, and improving the reliability of the detection results.

[0022] Optionally, the image optimization module is further configured with a light source optimization strategy, including: Perform feature analysis on the edge image of the bimetallic steel strip to determine the edge image shooting area and match the lighting intensity corresponding to the optimal shooting lens in the preset lighting database; The light source of the edge image shooting area is adjusted according to the light intensity and the preset parallel illumination angle, and the light source adjustment gain value is calculated using the preset light source gain analysis model to keep the light source adjustment gain value within the preset reference gain value range.

[0023] By adopting the above technical solution, the light source optimization strategy matches the light intensity and adjusts the gain value according to the edge area, which helps to achieve precise adaptation of the light source to the lens and the edge characteristics of the steel strip, improves the uniformity of light and image clarity, avoids edge recognition errors caused by improper lighting, and maintains the stability of the light source adjustment effect.

[0024] Optionally, the light source gain analysis model is calculated using the following formula: ; in, is the light source gain adjustment value, is the grayscale standard deviation of the image after adjustment, Used to reflect the uniformity of lighting. is the preset grayscale standard deviation of the image before adjustment, is the edge position detection error, is the preset reference edge error coefficient, is the contrast improvement of the adjusted image, is the preset reference contrast.

[0025] By adopting the above technical solution, the light source gain analysis model quantitatively calculates the gain value G through a formula. Combined with grayscale uniformity, edge error and contrast parameters, it helps to accurately evaluate the light source adjustment effect, realize quantitative control of the gain value, avoid the subjectivity of light source adjustment, improve the accuracy and consistency of lighting adjustment, and ensure stable image quality.

[0026] In summary, this application includes at least one of the following beneficial technical effects: 1. The precision control, lens distortion calibration, and deviation adjustment modules work together, combined with edge image acquisition and processing, to help improve the straightness detection accuracy and calibration reliability of bimetallic steel strips. This system also adapts to the detection requirements of steel strips of different specifications, reduces secondary processing caused by calibration errors, and improves the first-pass rate and overall efficiency of steel strip shearing. 2. The camera position pre-calibration strategy generates position compensation coefficients through coordinate analysis and error distance calculation. This helps to eliminate camera installation position errors in advance, reduces interference from the second camera's position deviation on measurement, improves the baseline consistency of position value detection, and provides more reliable raw data for subsequent deviation calculations. 3. The lens distortion calibration strategy uses sub-pixel corner detection, polynomial distortion model correction, and image smoothing to help accurately eliminate edge pixel deviations caused by lens distortion, improve the accuracy of edge feature position recognition, ensure that straight line measurement errors are controlled within a preset range, and provide high-quality image data for deviation calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a schematic diagram of the system module connection in this application.

[0028] Figure 2 It is a method flow chart of steps S100 to S102 in this application.

[0029] Figure 3 It is a method flow chart of steps S200 to S201 in this application.

[0030] Figure 4 It is a method flow chart of steps S300 to S302 in this application.

[0031] Figure 5 It is a method flow chart of steps S400 to S402 in this application.

[0032] Figure 6 It is a method flow chart of steps S403 to S404 in this application.

[0033] Figure 7It is a method flow chart of steps S4041 to S4043 in this application.

[0034] Figure 8 It is a method flow chart of steps S500 to S501 in this application. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figures 1-8 It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0036] The embodiments of the present invention are described in further detail below with reference to the accompanying drawings.

[0037] The embodiment of the present application discloses a bimetallic steel strip straightness calibration and detection system based on three-camera group measurement. Through the mutual cooperation between multiple modules, the modules such as camera measurement precision control, lens distortion calibration and deviation adjustment can be completed. Combined with edge image acquisition and processing, it helps to improve the detection accuracy and calibration reliability of the straightness of the bimetallic steel strip, adapt to the detection requirements of steel strips of different specifications, enhance the stability of edge feature recognition, avoid the accuracy limitations of traditional mechanical calibration, reduce secondary processing caused by calibration errors, and thus improve the overall efficiency of steel strip shearing.

[0038] Reference Figure 1 The bimetallic steel strip linear calibration detection system based on three-camera group measurement includes the following modules: A camera measurement module is provided with a first camera, a second camera, and a third camera arranged at equal intervals along the running path of the bimetallic steel strip after shearing, for respectively capturing edge images of the bimetallic steel strip at three different positions; In the embodiment of this application, the first, second, and third cameras are industrial area array cameras, such as 12-megapixel CMOS cameras, spaced equidistantly horizontally along the strip's travel direction. Adjacent cameras are spaced 400 mm apart, with their lenses oriented perpendicularly to the strip's travel direction. This ensures complete capture of the strip's edge contour. The cameras are equipped with adjustable focus lenses with a focal length of 12-35 mm and a trigger-activated camera mechanism that synchronizes with the strip's travel speed, such as a 0.1-second interval at 1 m / s, to prevent image blur. For example, when a bimetallic strip travels at 0.8 m / s, the cameras can capture edge features at various locations in real time, providing raw image data for subsequent straightness calculations.

[0039] an accuracy control module connected to the camera measurement module and configured with a camera accuracy matching strategy for adjusting the detection accuracy of edge images captured by the first camera, the second camera, and the third camera; This module ensures that the detection accuracy of a single camera remains stable within ±0.01mm through hardware parameter adjustments such as exposure time, gain, and software algorithm compensation. The camera precision matching strategy helps to maintain consistency in the measurement errors of the three cameras, avoiding straightness calculation errors caused by single-camera precision deviations, thereby meeting the calibration requirement of a D-value straightness deviation within ±0.03mm. The specific precision matching strategy will be further disclosed in subsequent steps. For example, for bimetallic steel strips of varying thicknesses, the module can automatically adjust camera parameters to ensure that even subtle deformations at the edge of thin steel strips are accurately captured.

[0040] An image optimization module, connected to the camera measurement module, and configured with a lens distortion calibration strategy for performing distortion correction on the camera-captured image; This module uses multiple interchangeable industrial lenses adapted to different models for image capture and corrects the image using distortion correction algorithms, such as the Zhang calibration method, to eliminate the effects of radial and tangential lens distortion on edge measurement. Furthermore, the module utilizes a parallel light source, such as an LED strip light with adjustable brightness, to illuminate the edge area from the oblique bottom of the steel strip, enhancing the contrast between the edge and the background and reducing reflective interference. Lens distortion correction strategies help improve the authenticity of edge contours in images, providing a reliable image foundation for subsequent deviation calculations. The specific calibration methods will be further explained later. For example, for bimetallic steel strips with curved edges, after distortion correction, the straightness error of the edge in the image can be reduced to within 0.005mm.

[0041] The image processing module is connected to the camera measurement module and is configured with a deviation adjustment strategy for calculating and analyzing the straightness deviation value of the edge image to determine the shear deviation of the bimetallic steel strip and perform feedback adjustment.

[0042] After receiving edge images captured by three cameras, the module extracts the pixel coordinates of the steel strip's edges using an edge detection algorithm, such as the Canny operator, and converts them to actual physical dimensions in pixels-to-millimeter conversion ratios, for example, at a 1:0.001 ratio. Based on the measurement data from the three cameras, the widths measured by the first, second, and third cameras correspond to the edge positions X1, X2, and X3. Straightness deviation is calculated using the formula X1 + X3 / 2 - X2 to determine whether the steel strip deviates from a straight line. If the straightness deviation exceeds ±0.03mm, the module sends an adjustment signal, such as a motor drive command, to the PLC control system to control the calibration mechanism. In the embodiment, this is achieved by fine-tuning the steel strip using the pressure roller. This deviation adjustment strategy helps to correct for straightness deviations in the steel strip after shearing in real time. The specific adjustment logic will be further explained below. For example, if a straightness deviation of +0.04mm is detected, indicating that the steel strip is deviating to one side, the system can slightly apply pressure to the left pressure roller to correct the deviation to within 0.01mm.

[0043] Reference Figure 2 , the bias adjustment strategy includes the following steps: Step S100: performing feature recognition based on the edge image to determine position values ​​corresponding to edge feature positions of the bimetallic steel strip in the first, second, and third camera fields of view, the position values ​​including a first position value, a second position value, and a third position value; In step S100, the feature recognition of the edge image is mainly achieved through image processing algorithms. The system first pre-processes the original image captured by the camera, such as grayscale and denoising, and then uses edge detection operators such as the Sobel operator to extract the contour features of the edge of the bimetallic steel strip. The edge features are usually manifested as continuous lines with sudden changes in grayscale values ​​in the image, corresponding to the physical boundaries of the steel strip edge.

[0044] During execution, the system converts the pixel coordinates of the edge outline into actual physical position values ​​in millimeters. The edge position identified in the first camera's field of view is recorded as the first position value X1, the second camera's corresponding position value X2, and the third camera's corresponding position value X3. For example, for a 50mm wide bimetallic steel strip, if the first camera detects the edge at 20.12mm from the left baseline, then X1 = 20.12; the second camera detects 20.15mm, then X2 = 20.15; and the third camera detects 20.13mm, then X3 = 20.13.

[0045] This step converts image information into quantifiable physical position data, providing the basis for subsequent straightness deviation calculations. By accurately identifying edge feature locations, this ensures that the position values ​​truly reflect the actual operating state of the steel strip, for example, avoiding edge positioning errors caused by image blur.

[0046] The feature recognition of edge images adopts the combined algorithm of "preprocessing + adaptive edge detection". The specific execution process is as follows: Image preprocessing: The color image captured by the camera is first grayscaled. The RGB three-channel pixel values ​​are converted into single-channel grayscale values ​​using a weighted average algorithm. A median filter with a 3×3 filter kernel is then used to remove salt and pepper noise from the image. Dust and water droplets that may be present in the bimetallic steel strip production environment can cause noise in the image. Median filtering can effectively retain edge information while eliminating isolated noise points.

[0047] Adaptive edge detection: Using the improved Canny edge detection algorithm: First, smooth the image through Gaussian filtering to reduce the interference of high-frequency noise on edge extraction; Calculate the image gradient using the Sobel operator to obtain the horizontal and vertical gradients respectively, and determine the strength and direction of the edge; The non-maximum suppression algorithm is used to refine the edges, retaining only the local maximum pixels in the gradient direction and compressing the wide edges to a single pixel width; Based on the grayscale features of the steel strip edge, a dynamic double threshold is set, such as the high threshold T1 = image average grayscale × 1.2, and the low threshold T2 = T1 × 0.5. Strong edges exceeding T1 are screened out and weak edges between T1 and T2 and connected to the strong edges are connected, finally obtaining a continuous steel strip edge contour.

[0048] Sub-pixel positioning: Due to the high precision requirements on the edge of the steel strip, the corresponding camera detection accuracy is ±0.01mm. Based on the pixel-level edge, the grayscale distribution of pixels near the edge is fitted with a quadratic curve. The sub-pixel coordinate positioning accuracy corresponding to the grayscale value peak can reach 0.1 pixel level, and then converted into physical position values. For example, when the pixel size is 0.005mm / pixel, sub-pixel positioning can achieve a theoretical accuracy of ±0.0005mm.

[0049] For example, for a slightly reflective bimetallic steel strip image, the grayscale value difference in the edge area of ​​the preprocessed grayscale image is in the grayscale range of 50-800-255. The adaptive Canny algorithm can accurately extract the edge. After sub-pixel fitting, the errors of the obtained edge position values ​​X1, X2, and X3 can be controlled within ±0.005mm, providing reliable raw data for subsequent straightness deviation calculations.

[0050] This step converts image information into quantifiable physical location data. Algorithmic optimization ensures robustness and precision in edge feature recognition. Even with minor scratches and reflections on the steel strip surface, edge positions can be stably extracted, preventing subsequent calibration errors caused by feature recognition deviations.

[0051] Step S101: Calculating the straightness deviation value of the bimetallic steel strip based on the position value and a preset deviation calculation model, and comparing it with a preset standard straightness deviation range to determine the deviation result; In step S101, the deviation calculation model converts the first, second, and third position values ​​X1, X2, and X3 into a quantitative indicator of the steel strip's straightness—the straightness deviation value D. This allows the correlation between the three positions to determine whether the steel strip is running straight. The model's details will be further explained in subsequent steps.

[0052] During execution, the system first calls the deviation calculation model to process the position value and outputs a deviation value, D. This value is then compared to a preset standard straightness deviation range, such as ±0.03mm. If D is within this range, the deviation is considered acceptable. If D is outside this range, such as D = +0.04mm or D = -0.05mm, the deviation is considered excessive. For example, if D = 0.01mm calculated based on the position value in step S100 is within the ±0.03mm range, the deviation result is considered acceptable. However, if D = 0.04mm for another batch of steel strip, the result is considered excessive.

[0053] This step quantifies and compares the strip's straightness, providing a basis for subsequent adjustments. If the strip meets the standard, no intervention is required; if it exceeds the standard, an adjustment mechanism is triggered to ensure that the strip's running deviations are under control. For example, for bimetallic strips used in precision instruments, this step can quickly identify minor deviations, preventing product scrapping due to cumulative errors.

[0054] Step S102: Dynamically feedback-adjust the shearing position of the bimetallic steel strip according to the deviation result to keep the straightness deviation value within a preset standard straightness deviation range.

[0055] In step S102, dynamic feedback adjustment is implemented through a closed loop of "deviation result-adjustment instruction-executor": the system converts the deviation result into specific adjustment parameters, such as adjustment direction and adjustment amount, and sends them to the control module of the steel strip shearing equipment, such as the PLC system, and the shear position correction is completed by a mechanical actuator such as a guide roller driven by a servo motor.

[0056] During execution, if the deviation is "out of specification," such as D = +0.04mm, indicating the strip is deflecting to the right, the system calculates the adjustment, such as fine-tuning the left guide roller by 0.02mm, and sends a command to the actuator. After the adjustment, the camera recaptures the edge image, repeating steps S100-S101 until the deviation D returns to within the ±0.03mm range. For example, for a batch of steel strips, mechanical vibration resulted in D = 0.05mm. After the system actuated the adjustment mechanism, a retest revealed D = 0.02mm, returning the deviation to acceptable levels.

[0057] The purpose of this step is to achieve real-time response of "detection-judgment-adjustment" and to avoid the accumulation of deviations that affect the quality of the steel strip by dynamically correcting the shearing position.

[0058] It is necessary to further explain that the deviation calculation model uses the following formula: ; ; ; ; in, Indicates the straightness deviation value of the bimetallic steel strip, 、 、 Represent the first position value, the second position value and the third position value respectively, is the position compensation coefficient, is the distortion correction coefficient, is the light source influence coefficient, is the lateral deviation between the actual position of the second camera and the midpoint of the line connecting the first and third cameras, is the preset reference deviation value, is the preset position influence coefficient, is the difference between the actual deviation value of the edge pixel of the image and the theoretical distortion-free value after lens distortion calibration. is the preset reference distortion pixel value, is the distortion influence coefficient, Grayscale uniformity of the edge area of ​​the bimetallic steel strip under parallel light source illumination, is the preset lower limit of the effective uniformity of the light source, is the upper limit of the ideal uniformity of the light source, It is the preset light source influence coefficient.

[0059] Reference Figure 3 , the camera accuracy matching strategy includes: Step S200: Analyzing the straightness deviation value to determine a deviation range, and matching a lens model corresponding to the lens model in a preset accuracy database according to the deviation range, wherein the lens model has different shooting field angles; In step S200, the analysis of the deviation range is based on the difference between the straightness deviation value D and the standard range of ±0.03mm: if D is close to the upper limit of the standard range, such as a small deviation range, it means that the straightness fluctuation of the steel strip is gentle, and a higher-precision lens is required to capture subtle deviations; if D is far beyond the standard range, such as a large deviation range, the lens needs to cover a larger field of view to prevent the edge of the steel strip from exceeding the shooting range.

[0060] In the preset accuracy database, lens models are categorized by field of view and focal length: wide-field-of-view lenses are suitable for large deviation ranges, covering a larger area around the strip edge and preventing it from falling out of the frame; narrow-field-of-view lenses are suitable for smaller deviation ranges and have longer focal lengths, capturing edge details more clearly and improving positioning accuracy. During execution, the system matches the corresponding lens from the database based on the deviation range. For example, when the strip has just entered the production line and the deviation range is large, a wide-field-of-view lens is matched; once the deviation has stabilized within a small range, a narrow-field-of-view lens is automatically matched.

[0061] The purpose of this step is to find a balance between "complete edge capture" and "high-precision measurement" by adapting the lens model and deviation range, so as to avoid missed detection due to insufficient lens field of view or reduced local measurement accuracy due to excessive field of view.

[0062] Step S201: determining the optimal shooting lens for the bimetallic steel strip of current size specifications according to the corresponding lens models of the first camera, the second camera, and the third camera, and switching the shooting lens corresponding to the lens model based on the optimal shooting lens.

[0063] In step S201, the optimal camera lens is determined based on the current strip dimensions and the synergy between the three cameras, including factors such as width and thickness. For wide strips, the lens field of view must be sufficient to cover the entire width of the strip edge; for narrow strips, a higher-precision narrow-field-of-view lens can be used. Furthermore, the three cameras must use the same lens model to avoid measurement data deviations due to lens parameter differences.

[0064] During execution, the system first reads the steel strip's dimensions, such as those obtained through a pre-detection module. Combined with the lens model matched in step S200, it selects the optimal lens that meets both the "steel strip edge coverage" and "accuracy requirements." Automatic lens switching is then accomplished through a lens switching mechanism, such as a motorized lens turntable. After switching, the system performs a brief calibration, such as capturing an image of a standard plate, to ensure lens parameter stability. For example, when producing wide steel strip, the three cameras simultaneously switch to wide-field-of-view lenses; when switching to narrow strip production, they simultaneously switch to narrow-field-of-view, high-precision lenses.

[0065] The purpose of this step is to achieve dynamic adaptation of the lens to the steel strip specifications, while ensuring consistency in multi-camera measurement. That is, by unifying the lens model, deviations in X1, X2, and X3 measurements caused by differences in individual camera lenses can be avoided, providing a unified benchmark for subsequent straightness calculations.

[0066] Reference Figure 4 ,The camera measurement module is also configured with a camera position pre-calibration strategy, including: Step S300: establishing a standard coordinate axis orientation according to the moving direction of the bimetallic steel strip, performing coordinate analysis on the first camera and the third camera according to the standard coordinate axis orientation, and generating a reference line connecting the coordinates of the first camera and the second camera; The movement direction of the bimetallic steel strip refers to the transmission direction of the bimetallic steel strip on the production line, such as horizontally to the right, which is the basis for formulating coordinate reference.

[0067] The standard coordinate axis is a three-dimensional coordinate system established with the steel belt movement direction as the X-axis, the direction perpendicular to the steel belt surface as the Z-axis, and the horizontal direction perpendicular to the movement direction as the Y-axis. It is used to unify the position reference of the camera and the steel belt.

[0068] The baseline is a straight line connecting the Y-axis coordinates of the first and third cameras in the standard coordinate axis, corresponding to the position in the width direction of the steel strip. In theory, it is a reference line for judging whether the second camera is centered. Since the three cameras are arranged equidistantly, the second camera should be located directly above the midpoint of the baseline.

[0069] During implementation, the system first determines the positive X-axis direction based on the direction of the steel strip's motion, with the positive Z-axis direction perpendicular to the steel strip's surface and pointing upwards, establishing a standard coordinate axis. The system then uses the camera's built-in positioning sensor, such as a laser positioner, to obtain the Y-axis coordinates of the first and third cameras. The two coordinates are connected to create a baseline. For example, if the first camera's Y-axis coordinate is Y1 and the third camera's Y-axis coordinate is Y3, the baseline is the line connecting Y1 and Y3, which is used to subsequently determine the accuracy of the second camera's installation position.

[0070] The purpose of this step is to establish a unified coordinate reference system, clarify the position relationship between the first and third cameras through the baseline, and provide a "benchmark ruler" for the subsequent position error analysis of the second camera.

[0071] Step S301: performing point-line distance analysis based on the corresponding coordinates of the baseline and the second camera located in the middle to determine the error distance between the second camera and the baseline; The corresponding coordinates of the second camera refer to the actual measured coordinate Y2 of the second camera on the Y axis in the standard coordinate axis.

[0072] Point-line distance analysis: This method calculates the vertical distance from the second camera's coordinate point Y2 to the line connecting the first and third cameras on the baseline, and is used to quantify the degree of installation offset of the second camera.

[0073] Error distance: The result of point-line distance analysis, that is, the vertical distance between the actual position of the second camera and the reference line, reflecting the position deviation during installation, such as slight misalignment during mechanical installation.

[0074] During execution, the system uses a geometric algorithm based on the standard coordinate system to calculate the point-line distance formula. It compares the second camera's Y-axis coordinate, Y2, with the baseline to calculate the error distance. For example, if the baseline is a theoretically centered straight line, and the second camera's actual position deviates from this line due to installation deviation, the error distance is a measure of the degree of deviation.

[0075] The purpose of this step is to quantify the installation error of the second camera. Since it is difficult to completely achieve the theoretical center position in actual installation, the error distance can intuitively reflect the size of the deviation and provide a basis for subsequent compensation correction.

[0076] Step S302 : generating a position compensation coefficient based on the error distance and a preset position compensation model, and correcting the second position value detected by the second camera according to the position compensation coefficient.

[0077] Position compensation model: An algorithmic model used to convert error distance into compensation coefficient. The core of the model is to establish the corresponding relationship between "error distance-compensation amount". The specific logic will be expanded later.

[0078] Position compensation coefficient: A correction parameter calculated based on the error distance, used to adjust the position value detected by the second camera to offset the impact of installation errors.

[0079] Second position value correction: The position value originally detected by the second camera is affected by the installation error and combined with the compensation coefficient to obtain the true position value after eliminating the error.

[0080] During implementation, the system inputs the error distance into the position compensation model and generates a corresponding compensation coefficient. For example, the larger the error distance, the larger the absolute value of the compensation coefficient. This coefficient is then used to correct the second position value X2 detected by the second camera. For example, if the second camera is installed to the left, resulting in a smaller detection value, the compensation coefficient will adjust the corrected X2 to the right, closer to the true position.

[0081] The purpose of this step is to eliminate the influence of installation error on measurement through compensation correction, ensure that the position value of the second camera can truly reflect the edge status of the steel strip, avoid inaccurate calculation of straightness deviation value D due to camera position deviation, and ultimately improve the overall calibration accuracy.

[0082] Reference Figure 5 , the lens distortion calibration strategy includes: Step S400: performing sub-pixel corner detection based on the captured edge image to determine lens distortion parameters of the first camera, the second camera, and the third camera; Sub-pixel corner detection is a feature location method with higher accuracy than pixel-level detection. By fitting the grayscale distribution of pixels surrounding a corner, the corner's position is located within the pixel rather than at integer pixel coordinates, achieving micron-level positioning accuracy. Lens distortion parameters describe the degree to which the edge of a lens image deviates from an ideal straight line. These parameters include radial distortion, such as convex or concave edges, and tangential distortion, such as offset caused by tilt of the imaging plane. These parameters are determined by the optical properties of the lens.

[0083] In practice, the system first captures an image of a calibration plate with a standard checkerboard pattern. The checkerboard corners are precisely located feature points. Sub-pixel corner detection is then used to extract the actual coordinates of the corners in the image. This coordinate is then compared with the theoretical coordinates of the corners on the calibration plate, and the deviation between the two is calculated. This deviation represents a quantitative representation of lens distortion, which is then used to determine the distortion parameters. For example, ordinary industrial lenses may exhibit slight radial distortion, causing straight lines in edge images to appear slightly bent. This process accurately captures this distortion characteristic.

[0084] The purpose of this step is to provide "raw data" for distortion correction. By detecting and quantifying the distortion parameters, the deviation law of lens imaging can be clarified, laying the foundation for subsequent image correction.

[0085] Step S401: constructing a polynomial distortion model based on the distortion parameters, performing reverse coordinate mapping on the edge image of the bimetallic steel strip acquired in real time, and calculating a pixel-level distortion compensation value; The polynomial distortion model is a mathematical model that uses a polynomial function to fit the lens distortion pattern. By substituting the distortion parameters into a polynomial, such as a quadratic or cubic polynomial, a correspondence between ideal coordinates and distorted coordinates is established. Inverse coordinate mapping is the process of reverse calculation based on the distortion model. Given the coordinates of a pixel in a distorted image, the model is used to derive its corresponding coordinates in the ideal, undistorted image, achieving reverse correction of the coordinates. The pixel-level distortion compensation value is the value used to correct the position of a single pixel. It is the difference between the distorted pixel coordinates and the ideal coordinates, used to "pull" the pixel in the distorted image back to its correct position.

[0086] During execution, the system inputs the distortion parameters determined in step S400 into the polynomial distortion model to generate a distortion correction formula. When capturing a real-time image of the steel strip edge, the coordinates of each pixel in the image are inversely mapped and calculated to obtain a corresponding compensation value. For example, if a pixel deviates by 0.5 pixels from its ideal position due to radial distortion, the compensation value is -0.5 pixels, which is used to correct it to the correct position.

[0087] The purpose of this step is to convert the distortion parameters into an executable correction plan. Through mathematical models and coordinate mapping, the compensation amount for each pixel is quantified, providing a calculation basis for accurate correction of the image.

[0088] Step S402: Dynamically correct the camera field of view angle and edge feature position according to the distortion compensation value, keep the compensated straight line measurement error less than the preset upper limit deviation value, and smooth the corrected image to maintain the edge feature clarity.

[0089] The camera's field of view is the range of angles that the camera can capture. Distortion can cause the actual field of view to deviate from the theoretical value. Dynamic correction adjusts the calculation parameters of the field of view based on the compensation value to ensure that the imaging range is consistent with the actual value. The edge feature position is the coordinate of the steel strip edge in the image. Dynamic correction adjusts this coordinate using the compensation value to eliminate the position offset caused by distortion. The straight line measurement error is the deviation between the steel strip edge line in the corrected image and the ideal straight line. The upper limit deviation value is the maximum error allowed by the system, such as ±0.01mm. Smoothing is a noise reduction process on the corrected image, such as Gaussian blur, to reduce pixel jumps that may occur during the correction process while retaining the clear outline of the edge.

[0090] During execution, the system uses the compensation value to adjust the edge feature position coordinates and field of view parameters, then calculates the corrected linear measurement error. If the error exceeds the upper limit deviation value, the compensation value is re-optimized. If the error meets the standard, the image is smoothed to remove residual noise from the correction. For example, after correction, the linear measurement error of a steel strip edge is 0.008mm, which is less than the upper limit of 0.01mm. The smoothed edge contour is continuous and has no jagged edges, making it directly usable for subsequent deviation calculations.

[0091] The purpose of this step is to control the impact of distortion on measurement within the allowable range through dynamic correction and optimization, while ensuring image quality and the accuracy of subsequent edge feature recognition and deviation calculation.

[0092] Reference Figure 6 The lens distortion calibration strategy also includes a lens field of view enhancement sub-strategy, which uses the following steps to optimize the shooting field of view angle of the camera measurement module: Step S403: matching the corresponding optimal lens field of view angle in the preset database according to the shearing parameters of the bimetallic steel strip, and adjusting the shooting field of view angles of the first camera, the second camera, and the third camera; During execution, the system first obtains the current strip's cutting parameters. Based on the correspondence between cutting parameters and field of view angles in a preset database, it matches the optimal lens field of view angle that fully covers the strip's edges while retaining a reasonable amount of redundancy. The lens's motorized adjustment components then synchronously adjust the three cameras' field of view angles to matching values—for example, for wide strips, a wider field of view angle is used to prevent the edges from exceeding the frame; for narrow strips, a smaller field of view angle is used to reduce background interference.

[0093] The purpose of this step is to ensure that the camera can always capture complete edge features by adapting the field of view angle to the steel strip specifications, providing effective original images for subsequent distortion calibration and edge recognition.

[0094] Step S404: Monitor the changes in the cropping parameters to determine the triggering of the field of view angle change instruction, and update the optimal field of view lens angle according to the changed cropping parameters, and perform image analysis based on the updated optimal field of view lens angle with a preset verification sub-strategy to keep the camera's image shooting size within a preset standard size range.

[0095] Updating the optimal field of view lens angle involves re-matching the corresponding field of view angle from the database based on the new cropping parameters. The pre-set verification sub-strategy is an analytical method used to verify the suitability of the adjusted field of view angle. This is verified by checking the integrity of the steel strip edge in the image and whether the distance from the edge to the edge of the frame meets the requirements, such as retaining at least 5% of the frame width. The image capture size refers to the pixel size of the steel strip area in the image. The standard size range is the size range that ensures edge recognition accuracy, such as the steel strip area occupying 60%-80% of the image width.

[0096] During specific execution, if the system detects that the width of the steel strip changes from 50mm to 70mm, with a change of 20mm greater than the threshold of 10mm, the system triggers a change instruction and updates the optimal field of view angle to 40°. After the adjustment, the image is analyzed through the verification sub-strategy. If the edge of the steel strip is intact and the redundant distance meets the standard, the adjustment is confirmed to be effective. If the edge part exceeds the screen, the field of view angle is re-fine-tuned until the image shooting size is within the standard range.

[0097] The purpose of this step is to achieve dynamic adaptation of the field of view angle. By monitoring parameter changes and verifying the adjustment effect, it ensures that the camera can still stably capture qualified edge images when the steel strip specifications are switched, providing a reliable image source for subsequent distortion calibration and deviation calculation.

[0098] Reference Figure 7 , the verification sub-strategy includes: Step S4041: determining a width pixel range, a height pixel range, and an edge feature clarity threshold of an optimal image according to preset standard image size parameters, wherein the standard image size parameters correspond to shearing parameters of the bimetallic steel strip; The standard image size parameter refers to the image size benchmark set to ensure edge recognition accuracy. It is predefined based on the shearing parameters of the bimetallic steel strip, such as width and thickness. For example, wide steel strips correspond to a larger width pixel range, while narrow steel strips correspond to a relatively small range. This ensures that the steel strip area in the image is neither compressed nor redundant. The width pixel range and height pixel range are specific manifestations of the standard image size parameters, respectively limiting the pixel count range in the horizontal and vertical directions of the image, such as a width pixel range of 800-1200 pixels and a height pixel range of 500-800 pixels. The edge feature clarity threshold is a benchmark value for measuring the clarity of edge contours, quantified by the edge gradient value. For example, the threshold is ≥80, and higher gradient values ​​indicate clearer edges.

[0099] During specific execution, the system calls the corresponding standard image size parameters based on the current steel strip cutting parameters, such as a width of 60mm, and determines that the optimal image must meet the requirements of 800-1000 pixels in width, 500-600 pixels in height, and an edge clarity threshold of ≥85, providing a judgment standard for subsequent image verification.

[0100] The purpose of this step is to establish the "qualified standard" for image quality, by clarifying the quantitative range of size and clarity, to ensure that subsequent verification has a clear basis.

[0101] Step S4042: Acquire the bimetallic steel strip image at the current viewing angle, and extract the actual width pixel value, actual height pixel value, and edge feature clarity value of the image; The actual width pixel value is the total number of pixels in the horizontal direction of the current captured image, such as 900 pixels. The actual height pixel value is the total number of pixels in the vertical direction, such as 550 pixels. Both reflect the actual size of the image. The edge feature clarity value is a quantitative value obtained by image algorithms, such as calculating the edge gradient intensity, such as 90. It is used to measure the clarity of the steel strip edge in the current image.

[0102] During specific execution, the system controls the camera to capture the steel strip image in the current field of view, and automatically extracts the above three actual values ​​through image processing tools, for example, the image width is 920 pixels, the height is 580 pixels, and the edge clarity value is 88, providing actual data for subsequent comparison.

[0103] The purpose of this step is to obtain the actual quality parameters of the current image as a basis for comparison with the standard parameters.

[0104] Step S4043: Based on the comparison between the actual size value and the standard size parameters, the size deviation value and the clarity deviation value are calculated. If both are within the preset allowable deviation range, it is determined that the current field of view angle meets the requirements; if they exceed, the camera field of view angle is fine-tuned according to the deviation value and the image is recaptured until the image size and clarity meet the standard parameter requirements.

[0105] The size deviation is the difference between the actual width / height pixel values ​​and the standard range. For example, if the actual width is 920 pixels and the standard range is 800-1000 pixels, the deviation is 20 pixels within the acceptable range. The clarity deviation is the difference between the actual edge feature clarity value and the threshold, for example, if the actual value is 88 and the threshold is 85, the deviation is 3. The allowable deviation range is a preset acceptable deviation range, such as size deviation ≤ ±100 pixels and clarity deviation ≥ -5, to ensure that image quality remains within a reasonable fluctuation range.

[0106] During specific execution, the system compares the actual value with the standard parameters: if the size deviation and clarity deviation are both within the allowable range, the current field of view angle is determined to be qualified; if the actual width of 1300 pixels exceeds the standard upper limit of 100 pixels, the deviation value of 200 is calculated, the lens is controlled to reduce the field of view angle, and the image is recaptured and verified until it meets the requirements.

[0107] The purpose of this step is to ensure that the camera field of view angle can always produce qualified images through quantitative comparison and dynamic adjustment, providing a high-quality image foundation for subsequent edge recognition and calibration.

[0108] Reference Figure 8 ,The image optimization module is also equipped with light source optimization strategies, including: Step S500: performing feature analysis on the edge image of the bimetallic steel strip to determine the edge image shooting area, and matching the illumination intensity corresponding to the optimal shooting lens in the preset illumination database; The edge image capture area refers to the core region of the image where the edge of the bimetallic steel strip is located, typically the edge and a certain range nearby, ensuring that the edge features are fully presented. Feature analysis can locate the pixel range of this area, such as a 50-pixel strip on each side of the edge. The preset lighting database stores lighting parameters for different cameras, corresponding to different field of view angles and adapted to the steel strip edge features. Light intensity is a core parameter—for example, a wide-field-of-view lens requires higher light intensity to cover a larger area, while a narrow-field-of-view lens can appropriately reduce the intensity to avoid reflections. The light intensity corresponding to the optimal camera lens is the light intensity value that creates a clear grayscale difference between the edge of the area and the background, facilitating edge recognition.

[0109] During specific execution, the system locates the edge shooting area from the edge image, combines the currently used optimal shooting lens model, and matches the corresponding light intensity from the preset lighting database. For example, for the edge area of ​​the narrow field of view lens, it matches the light intensity of 500 lux to provide initial parameters for light source adjustment.

[0110] The purpose of this step is to provide a suitable lighting basis for the edge area by matching the adaptive light intensity, so as to avoid blurred edges due to too dark light or reflections due to too bright light, which will affect subsequent feature recognition.

[0111] Step S501: adjusting the light source of the edge image shooting area according to the light intensity and the preset parallel illumination angle, and calculating the light source adjustment gain value using the preset light source gain analysis model to keep the light source adjustment gain value within the preset reference gain value range.

[0112] The parallel illumination angle is set at 45° between the light source and the surface of the bimetallic strip. This reduces specular reflections caused by vertical illumination. A mechanical structure maintains this angle, ensuring uniform illumination of the edge capture area. The light source gain adjustment value is used to fine-tune the actual light source output intensity. Calculated based on the initial light intensity, it compensates for changes in ambient light or variations in reflectivity on the strip surface. The baseline gain value range is a preset, normal gain range, such as 0.8-1.2, ensuring that light source intensity is adjusted within a reasonable range to avoid image quality fluctuations caused by over-adjustment.

[0113] During specific execution, the system first turns on the light source according to the matched light intensity and preset parallel angle, and then calculates the currently required adjustment gain value through the light source gain analysis model; if the gain value is within the reference range, such as 1.0, the light source state is maintained; if the gain value drops to 0.7 below the lower limit due to increased ambient light, the light source intensity is appropriately increased to return the gain value to the range.

[0114] The purpose of this step is to offset the impact of external interference on lighting by dynamically adjusting the intensity of the light source, ensuring that the light in the edge shooting area is always stable, and providing continuous guarantee for the clear presentation of edge features.

[0115] It should be further explained that the light source gain analysis model is calculated using the following formula: ; in, is the light source gain adjustment value, is the grayscale standard deviation of the image after adjustment, Used to reflect the uniformity of lighting. is the preset grayscale standard deviation of the image before adjustment, is the edge position detection error, is the preset reference edge error coefficient, is the contrast improvement of the adjusted image, is the preset reference contrast.

[0116] Those skilled in the art will clearly understand that for the sake of convenience and brevity, the division of the above-mentioned functional modules is only used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working processes of the above-mentioned systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0117] An embodiment of the present invention provides a computer-readable storage medium storing a computer program capable of being loaded and executed by a processor for a bimetallic steel strip linear calibration and detection system based on three-camera group measurement.

[0118] Computer storage media include, for example, various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0119] Those skilled in the art will clearly understand that for the sake of convenience and brevity, the division of the above-mentioned functional modules is only used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working processes of the above-mentioned systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0120] The above are all preferred embodiments of the present application and are not intended to limit the scope of protection of this application. Unless otherwise stated, any feature disclosed in this specification, including the abstract and drawings, may be replaced by other equivalent or similar features. In other words, unless otherwise stated, each feature is merely an example of a series of equivalent or similar features.

Claims

1. A bimetallic steel strip linear calibration detection system based on three-camera group measurement, characterized in that: include: A camera measurement module is provided with a first camera, a second camera, and a third camera arranged at equal intervals along the running path of the bimetallic steel strip after shearing, for respectively capturing edge images of the bimetallic steel strip at three different positions; an accuracy control module connected to the camera measurement module and configured with a camera accuracy matching strategy for adjusting the detection accuracy of edge images captured by the first camera, the second camera, and the third camera; An image optimization module, connected to the camera measurement module, and configured with a lens distortion calibration strategy for performing distortion correction on the camera-captured image; The image processing module is connected to the camera measurement module and is configured with a deviation adjustment strategy for calculating and analyzing the straightness deviation value of the edge image to determine the shear deviation of the bimetallic steel strip and perform feedback adjustment.

2. The bimetallic steel strip linear calibration and detection system based on three-camera group measurement according to claim 1 is characterized in that: The deviation adjustment strategy includes: Performing feature recognition based on the edge image to determine position values ​​corresponding to edge feature positions of the bimetallic steel strip in the first, second, and third camera fields of view, the position values ​​including a first position value, a second position value, and a third position value; The straightness deviation value of the bimetallic steel strip is calculated based on the position value and the preset deviation calculation model, and compared with the preset standard straightness deviation range to determine the deviation result; The shearing position of the bimetallic steel strip is dynamically feedback-adjusted according to the deviation result to keep the straightness deviation value within the preset standard straightness deviation range.

3. The bimetallic steel strip linear calibration and detection system based on three-camera group measurement according to claim 2 is characterized in that: The deviation calculation model uses the following formula: ； ； ； ； in, Indicates the straightness deviation value of the bimetallic steel strip, 、 、 Represent the first position value, the second position value and the third position value respectively, is the position compensation coefficient, is the distortion correction coefficient, is the light source influence coefficient, is the lateral deviation between the actual position of the second camera and the midpoint of the line connecting the first and third cameras, is the preset reference deviation value, is the preset position influence coefficient, is the difference between the actual deviation value of the edge pixel of the image and the theoretical distortion-free value after lens distortion calibration. is the preset reference distortion pixel value, is the distortion influence coefficient, Grayscale uniformity of the edge area of ​​the bimetallic steel strip under parallel light source illumination, is the preset lower limit of the effective uniformity of the light source, is the upper limit of the ideal uniformity of the light source, It is the preset light source influence coefficient.

4. The bimetallic steel strip linear calibration and detection system based on three-camera group measurement according to claim 2 is characterized in that: The camera accuracy matching strategy includes: Analyzing the straightness deviation value to determine a deviation range, and matching a corresponding lens model in a preset accuracy database according to the deviation range, wherein the lens model has different shooting field angles; The optimal shooting lens of the bimetallic steel strip of the current size specification is determined according to the corresponding lens models of the first camera, the second camera and the third camera, and the shooting lens corresponding to the lens model is switched based on the optimal shooting lens.

5. A bimetallic steel strip straight line calibration and detection system based on three-camera group measurement according to claim 1 or 3, characterized in that: The camera measurement module is also configured with a camera position pre-calibration strategy, including: A standard coordinate axis orientation is established according to the moving direction of the bimetallic steel strip, and coordinate analysis is performed on the first camera and the third camera according to the standard coordinate axis orientation, and a reference line connecting the coordinates of the first camera and the second camera is generated; Perform point-line distance analysis based on the corresponding coordinates of the baseline and the second camera located in the middle to determine the error distance between the second camera and the baseline; A position compensation coefficient is generated based on the error distance and a preset position compensation model, and the second position value detected by the second camera is corrected according to the position compensation coefficient.

6. The bimetallic steel strip linear calibration and detection system based on three-camera group measurement according to claim 1 is characterized in that: The lens distortion calibration strategy includes: performing sub-pixel corner detection based on the captured edge image to determine lens distortion parameters of the first camera, the second camera, and the third camera; A polynomial distortion model is constructed based on the distortion parameters, reverse coordinate mapping is performed on the edge image of the bimetallic steel strip acquired in real time, and a pixel-level distortion compensation value is calculated; The camera field of view angle and edge feature position are dynamically corrected according to the distortion compensation value, so as to keep the compensated straight line measurement error less than the preset upper limit deviation value, and the corrected image is smoothed to maintain the edge feature clarity.

7. The bimetallic steel strip linear calibration and detection system based on three-camera group measurement according to claim 6 is characterized in that: It also includes a lens field of view enhancement sub-strategy, which uses the following steps to optimize the shooting field of view angle of the camera measurement module: Matching the corresponding optimal lens field of view angle in the preset database according to the shearing parameters of the bimetallic steel strip, and adjusting the shooting field of view angles of the first camera, the second camera, and the third camera; Monitor the changes in shearing parameters to determine the triggering of the field of view angle change instruction, and update the optimal field of view lens angle according to the changed shearing parameters. Perform image analysis based on the updated optimal field of view lens angle with a preset verification sub-strategy to keep the camera's image capture size within a preset standard size range.

8. The bimetallic steel strip linear calibration and detection system based on three-camera group measurement according to claim 7 is characterized in that: The verification sub-strategy includes: Determining a width pixel range, a height pixel range, and an edge feature clarity threshold of an optimal image based on preset standard image size parameters, wherein the standard image size parameters correspond to shearing parameters of the bimetallic steel strip; Collect the bimetallic steel strip image at the current field of view angle, and extract the actual width pixel value, actual height pixel value and edge feature clarity value of the image; Based on the comparison between the actual size value and the standard size parameters, the size deviation value and clarity deviation value are calculated. If both are within the preset allowable deviation range, it is determined that the current field of view angle meets the requirements; if they exceed, the camera field of view angle is fine-tuned according to the deviation value and the image is recaptured until the image size and clarity meet the standard parameter requirements.

9. The bimetallic steel strip linear calibration and detection system based on three-camera group measurement according to claim 4 is characterized in that: The image optimization module is also configured with a light source optimization strategy, including: Perform feature analysis on the edge image of the bimetallic steel strip to determine the edge image shooting area and match the lighting intensity corresponding to the optimal shooting lens in the preset lighting database; The light source of the edge image shooting area is adjusted according to the light intensity and the preset parallel illumination angle, and the light source adjustment gain value is calculated using the preset light source gain analysis model to keep the light source adjustment gain value within the preset reference gain value range.

10. The bimetallic steel strip linear calibration and detection system based on three-camera group measurement according to claim 9 is characterized in that: The light source gain analysis model is calculated using the following formula: ； in, is the light source gain adjustment value, is the grayscale standard deviation of the image after adjustment, Used to reflect the uniformity of lighting. is the preset grayscale standard deviation of the image before adjustment, is the edge position detection error, is the preset reference edge error coefficient, is the contrast improvement of the adjusted image, is the preset reference contrast.

Images