Laser cutting equipment control method and system for flexible OLED screen

Abstract

The invention relates to the technical field of flexible OLED screen manufacturing, in particular to a laser cutting equipment control method and system for a flexible OLED screen, and the method comprises the steps: obtaining an original image of the flexible OLED screen, removing reflection and noise to obtain an enhanced image, and extracting the edge contour and geometric parameters of a functional area of the flexible OLED screen; and when the parameters exceed the limit, adjusting the optical parameters, re-extracting the edge feature point sequence of the functional area, generating an initial cutting path curve, detecting the matching degree and smoothness of the path curve, correcting the offset area, if the update state of the offset area of the adjusted cutting path curve meets the standard, obtaining a qualified cutting path curve, and then executing the cutting operation. And obtaining a flexible OLED screen finished product. And evaluating the edge quality of the finished screen, carrying out electrical and optical performance tests on the finished screen, and carrying out adjustment and feedback to obtain a stable and matched final path. According to the method, the technical problem that the cutting precision of the flexible OLED screen is insufficient is solved.

Description

Technical Field

[0001] This invention relates to the field of flexible OLED screen manufacturing technology, and in particular to a control method and system for laser cutting equipment used in flexible OLED screens. Background Technology

[0002] With the development of modern display technology, flexible OLED screens have gained a significant position in consumer electronics products such as smartphones and wearable devices due to their high resolution, thinness, flexibility, and low power consumption. Laser cutting equipment for flexible OLED screens is widely used in factory control systems for producing finished flexible OLED screens. As market demand for diverse screen sizes and complex shapes increases, laser cutting equipment control technology has become a key process for achieving high-precision screen processing, and its processing method directly determines the quality of the final product.

[0003] Under current technology, when processing flexible OLED screens, laser cutting equipment control technology mainly relies on traditional image processing and fixed parameter cutting path planning. It is difficult to dynamically capture the irregular curves of functional area edges caused by material characteristics or production processes, resulting in cutting path deviations. At the same time, the existing technology has insufficient ability to adjust the optical system in dynamic environments, making it difficult to optimize the focal length and illumination intensity based on real-time image quality. This results in inaccurate detection of functional area edges, which can easily damage the pixel parts of functional areas during cutting and positioning, thereby affecting the screen yield.

[0004] Therefore, there is a problem that the laser cutting equipment control technology cannot accurately extract the edge contour of the functional area of ​​the flexible OLED screen and generate the optimal cutting path to match it. As a result, when there are non-standard curves or ambient light interference at the edge of the screen, the cutting equipment cannot adapt quickly, resulting in pixel or size deviation of the functional area during cutting. This seriously affects the quality and production efficiency of the flexible OLED screen, which directly restricts the competitiveness of the flexible OLED screen in the high-end market and fails to improve the intelligence level and production consistency of the laser cutting equipment.

[0005] In summary, existing technologies suffer from insufficient cutting precision in flexible OLED screens. Summary of the Invention

[0006] This invention provides a control method and system for laser cutting equipment for flexible OLED screens, which can solve the technical problem of insufficient cutting precision of flexible OLED screens in the prior art.

[0007] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a control method for laser cutting equipment for flexible OLED screens, comprising: The original image data of the flexible OLED screen is acquired, and light reflection and noise signals are removed from the original image data to obtain an enhanced image; Based on the feature point distribution of the enhanced image, the edge contours of the functional areas of the flexible OLED screen are extracted, and the geometric parameters of the edge contours of the functional areas are determined. If the geometric parameters of the functional area edge contour deviate from the preset parameter threshold by more than the allowable range, the focal length and illumination intensity parameters of the optical system are adjusted to obtain optimized image quality indicators. Based on the optimized image quality index, the functional area edge feature point sequence is re-extracted to obtain an accurate set of functional area edge point coordinates. The precise set of functional area edge point coordinates is sorted and transformed to generate an initial cutting path curve. The matching degree between the initial cutting path curve and the functional area edge is calculated, and the path offset area in the initial cutting path curve is obtained. Calculate the smoothness of the initial cutting path curve. If the smoothness is lower than a preset smoothness threshold, adjust the path offset region. If the update status of the path offset region of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained. Perform the cutting operation according to the qualified cutting path curve to obtain the finished flexible OLED screen; Based on the finished flexible OLED screen, the edge quality of the finished screen is evaluated, and the electrical and optical performance of the finished screen is tested. Through parameter adjustment and real-time feedback adjustment, a stable matching final path is obtained.

[0008] In one optional implementation, based on the feature point distribution of the enhanced image, the functional area edge contours of the flexible OLED screen are extracted, and the geometric parameters of the functional area edge contours are determined, including: Based on the feature point distribution in the enhanced image, Hough transform is used to detect straight line and curve segments to obtain a preliminary set of boundary points. If there are small discontinuous regions in the preliminary set of boundary points, then Bézier curve fitting is used to connect the discontinuous points to obtain a continuous edge contour candidate set. The ideal functional area outline in the CAD design drawing is used as a template, and shape matching is performed with the candidate set of continuous edge outlines. The outline with the highest matching degree is the continuous edge outline of the functional area. By performing polynomial fitting on the continuous edge contour of the functional area, the curvature and length are calculated to obtain a set of geometric parameters. Based on the set of geometric parameters, a data format conversion is performed to convert the parameter set into a standardized contour description file, thereby obtaining the geometric parameters of the functional area edge contour.

[0009] In one optional implementation, if the geometric parameters of the functional area edge contour deviate from a preset parameter threshold beyond the allowable range, the focal length and illumination intensity parameters of the optical system are adjusted to obtain optimized image quality indicators, including: The geometric parameters of the edge contour of the functional area are detected. If the deviation between the detection result and the preset parameter threshold exceeds the allowable range, the focal length parameter of the optical system is iteratively adjusted to obtain the adjusted focal length value. Based on the adjusted focal length value, the illumination intensity parameters of the optical system are optimized using light intensity equalization to obtain an optimized illumination intensity value; By using the adjusted focal length value and the optimized illumination intensity value, the enhanced image is re-acquired, resulting in an optimized image quality index.

[0010] In one optional implementation, based on the optimized image quality index, the functional area edge feature point sequence is re-extracted to obtain a precise set of functional area edge point coordinates, including: Based on the optimized image quality index, the edge feature point sequence is extracted again from the enhanced image using Hough transform to obtain a preliminary edge point dataset. Contour fitting and matching filtering are then performed to obtain a preliminary edge point dataset for the functional area. If the geometric parameters of the preliminary edge point dataset of the functional area exceed the preset geometric parameter threshold, the edge point coordinates are adjusted to obtain an accurate set of functional area edge point coordinates.

[0011] In one optional implementation, the precise set of functional area edge point coordinates is sorted and transformed to generate an initial cutting path curve. The matching degree between the initial cutting path curve and the functional area edge is calculated, and the path offset region in the initial cutting path curve is obtained, including: Based on the set of coordinates of the edge points of the functional area, the point sequence is sorted to obtain an ordered set of point sequences; For the ordered set of points, an affine transformation is used to perform coordinate transformation to obtain the transformed set of points. If the smoothness parameter of the transformed point sequence set deviates from the preset smoothness threshold by more than the allowable range, then the curve trajectory is adjusted by Bézier curve fitting and a certain distance of overall offset is made to generate the initial cutting path curve. Based on the initial cutting path curve, the smoothness and the matching degree of the functional area edge are detected by boundary comparison to obtain the path offset area.

[0012] In one optional implementation, the smoothness of the initial cutting path curve is calculated. If the smoothness is lower than a preset smoothness threshold, the path offset region is adjusted. If the updated state of the path offset region of the adjusted cutting path curve meets a preset standard, a qualified cutting path curve is obtained, including: Obtain the adjusted cutting path curve, extract curvature distribution features, and obtain the curvature distribution set; If there are points in the curvature distribution set whose curvature values ​​exceed a preset threshold, then vector correction is used to adjust the coordinates of the corresponding points to obtain a set of correction vectors; Using the set of correction vectors, the point sequence set is updated by coordinate transformation to obtain the updated point sequence set; For the updated set of point sequences, edge detection is used to calculate the matching degree with the edge of the functional area, and the path offset region update status of the adjusted cutting path curve is determined. If the path offset region update status of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained.

[0013] In one optional implementation, obtaining the adjusted cutting path curve includes: Based on the initial cutting path curve, the path smoothness parameter is calculated using the least squares method to obtain the set of deviation point coordinates; For the set of deviation point coordinates, gradient descent iteration is used to identify deviation points and obtain a set of correction point weights; Based on the set of correction point weights, an affine transformation is used to adjust the coordinates of the deviation points to obtain a set of adjusted point sequences. For the adjusted point sequence set, the matching degree between the boundary comparison detection and the functional area edge is obtained to obtain the adjusted cutting path curve.

[0014] In one optional implementation, a cutting operation is performed according to the qualified cutting path curve to obtain a finished flexible OLED screen, including: The qualified cutting path curve is smoothed by path optimization, and the point coordinates are adjusted by interpolation calculation to obtain a smoothed point sequence set. Based on the smoothed point sequence set, a control command sequence recognizable by the laser device is generated using path mapping. The deviation is corrected by coordinate transformation to obtain the corrected control command set. For the corrected set of control commands, the dynamic parameters of the laser device during operation are monitored in real time. If the dynamic parameters deviate from the preset threshold, the command sequence is adjusted and updated through feedback to obtain the updated command sequence. The updated instruction sequence is transmitted to a preset laser device via a device driver interface, which then controls the preset laser device to perform a cutting operation, resulting in a finished flexible OLED screen.

[0015] In one optional implementation, based on the finished flexible OLED screen, the edge quality of the finished screen is evaluated, and the electrical and optical performance of the finished screen is tested. Through parameter adjustment and real-time feedback adjustment, a stable matching final path is obtained, including: Based on the finished flexible OLED screen, the edge contour curve of the finished screen is extracted by image segmentation to evaluate the edge quality of the finished screen. Optical and electrical tests were conducted on the finished screen to obtain the corresponding test results; Based on the edge quality and test results, the laser cutting path is adjusted according to feedback to obtain a stable and matched final path.

[0016] In a second aspect, the present invention provides a control system for a laser cutting equipment for flexible OLED screens, comprising: The image acquisition module is used to acquire the original image data of the flexible OLED screen, remove light reflection and noise signals from the original image data, and obtain an enhanced image; The image contour feature extraction module is used to extract the edge contour of the functional area of ​​the flexible OLED screen based on the feature point distribution of the enhanced image, and determine the geometric parameters of the edge contour of the functional area. The image optimization module is used to adjust the focal length and illumination intensity parameters of the optical system to obtain optimized image quality indicators if the geometric parameters of the edge contour of the functional area deviate from the preset parameter threshold by more than the allowable range. The precise contour feature extraction module is used to re-extract the functional area edge feature point sequence based on the optimized image quality index, so as to obtain a precise set of functional area edge point coordinates; The initial cutting path generation module is used to sort and transform the coordinates of the precise functional area edge points, generate an initial cutting path curve, calculate the matching degree between the initial cutting path curve and the functional area edge, and obtain the path offset area in the initial cutting path curve. A qualified cutting path generation module is used to calculate the smoothness of the initial cutting path curve. If the smoothness is lower than a preset smoothness threshold, the path offset region is adjusted. If the update status of the path offset region of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained. The screen cutting module is used to perform a cutting operation according to the qualified cutting path curve to obtain a finished flexible OLED screen. The final cutting path generation module is used to evaluate the edge quality of the finished flexible OLED screen and conduct electrical and optical performance tests on the finished screen. Through parameter adjustment and real-time feedback adjustment, a stable and matched final path is obtained.

[0017] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention obtains the original image of the flexible OLED screen, removes reflection and noise to obtain an enhanced image, and extracts the edge contour and geometric parameters of the functional area of ​​the flexible OLED screen. When the parameters exceed the limit, the focal length and illumination intensity of the optical system can be adjusted in a targeted manner. Through multi-level enhancement processing of the original image, the adverse effects on the edge recognition accuracy are effectively eliminated. This not only ensures that the set of coordinates of the extracted functional area edge points has high-precision positioning, but also provides a reliable data basis for the subsequent planning of the laser cutting path.

[0018] (2) This invention generates an initial cutting path curve by re-extracting the functional area edge feature point sequence from the re-acquired enhanced image, detecting the matching degree and smoothness of the path curve, and correcting the path offset region. If the update state of the path offset region of the adjusted cutting path curve meets the standard, a qualified cutting path curve is obtained. The cutting operation is performed according to the qualified cutting path curve to obtain the flexible OLED screen finished product. This method ensures that the final output qualified cutting path curve matches the actual functional area edge contour through multiple rounds of detection and correction, and also has good smoothness. It can be directly used for actual cutting operations, effectively improving cutting efficiency and finished product qualification rate.

[0019] (3) This invention evaluates the edge quality of the finished screen and conducts electrical and optical performance tests on the finished screen. After adjustment and feedback, a stable matching final path is obtained. By dynamically optimizing the path through real-time monitoring and feedback, it adapts to variables in production and obtains a stable path. The stable path set can be used as a standardized template for cutting similar products, eliminating the need to perform image processing and path generation from scratch each time. This significantly improves the reusability of the process and cutting efficiency, and reduces time loss in mass production. Attached Figure Description

[0020] Figure 1 This is a schematic flowchart of a laser cutting equipment control method for flexible OLED screens provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of a control system for a laser cutting equipment used in flexible OLED screens, provided in the second embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figure 1 The first embodiment of the present invention provides a schematic diagram of a laser cutting equipment control method for flexible OLED screens, including steps S101 to S108, as follows: S101, acquire the original image data of the flexible OLED screen, remove light reflection and noise signals from the original image data, and obtain an enhanced image; S102, Based on the feature point distribution of the enhanced image, extract the edge contour of the functional area of ​​the flexible OLED screen and determine the geometric parameters of the edge contour of the functional area. S103, if the geometric parameters of the functional area edge contour deviate from the preset parameter threshold beyond the allowable range, then adjust the focal length and illumination intensity parameters of the optical system to obtain optimized image quality indicators; S104, Based on the optimized image quality index, re-extract the functional area edge feature point sequence to obtain an accurate set of functional area edge point coordinates; S105, sort and transform the precise set of functional area edge point coordinates to generate an initial cutting path curve, calculate the matching degree between the initial cutting path curve and the functional area edge, and obtain the path offset area in the initial cutting path curve. S106, calculate the smoothness of the initial cutting path curve. If the smoothness is lower than a preset smoothness threshold, adjust the path offset region. If the update status of the path offset region of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained. S107, Perform the cutting operation according to the qualified cutting path curve to obtain the finished flexible OLED screen; S108: Based on the finished flexible OLED screen, evaluate the edge quality of the finished screen and conduct electrical and optical performance tests on the finished screen. Through parameter adjustment and real-time feedback adjustment, obtain a stable matching final path.

[0023] In step S101, the original image data of the flexible OLED screen is acquired, and light reflection and noise signals in the original image data are removed to obtain an enhanced image.

[0024] First, a high-resolution sensor is used to capture images of the OLED screen surface. The resolution typically needs to be 1080p or higher to capture pixel-level details. For example, in the inspection of a curved-screen phone, a high-resolution sensor scans at 1920×1080 resolution to acquire raw RGB image data containing the screen's display content. This high resolution ensures that subsequent processing can accurately identify screen defects or display anomalies. After acquiring the images, data integrity must be ensured to avoid distortion caused by the screen's curvature.

[0025] It's worth noting that the acquired images require grayscale conversion and Gaussian filtering. Grayscale conversion transforms the RGB image into a single-channel grayscale image, reducing subsequent computational complexity. Gaussian filtering smooths the image and eliminates noise. For example, RGB values ​​are converted to grayscale values ​​using a weighted average method, with a common weight of R×0.299+G×0.587+B×0.114. Gaussian filtering further smooths the image and eliminates noise. For instance, a 5×5 Gaussian kernel with a standard deviation σ=1.5 is used to convolve the grayscale image, smoothing abrupt changes between pixels. This processing effectively reduces interference from minor scratches or dust on the screen surface, improving the accuracy of subsequent segmentation.

[0026] It's worth noting that in factory inspection scenarios, screens may reflect light under strong light, affecting image analysis. Therefore, reflective area separation is necessary. For example, reflective areas can be separated using local adaptive thresholding. Specifically, the threshold is dynamically determined by calculating the mean and variance of each pixel's neighborhood in the image. For instance, the local mean is calculated using an 11×11 pixel window, and the threshold is set to the mean plus an offset, such as 10, to separate the bright areas. This method is more suitable for handling screen images with uneven lighting compared to global thresholding.

[0027] It's worth noting that for images processed as described above, the Canny algorithm highlights edge features through gradient calculation, double thresholding, and edge connection. Specifically, the Canny algorithm first calculates gradients, typically using the Sobel operator, to calculate the gradients in the x and y directions, obtaining the gradient magnitude and direction for each pixel. Then, it uses non-maximum suppression to filter the edges, comparing the gradient magnitude of the current pixel with its two neighboring pixels along the gradient direction, retaining only the pixel with the largest gradient magnitude. This results in a set of candidate fine edges. To further filter effective edges, the algorithm sets two thresholds: a high threshold (H) and a low threshold (L) (typically H is 2-3 times L, e.g., H=150, L=50). If the gradient magnitude of a candidate point is greater than H, the point experiences a drastic change in grayscale and is determined to be a confirmed edge point. If the gradient magnitude is less than L, the point exhibits a slight change in grayscale and is determined to be noise or background, and is directly discarded. If the gradient magnitude is between H and L, it is determined to be an edge point to be determined. These points may have low gradient values ​​due to noise or gradual grayscale changes, or they may simply be noise, requiring verification in the edge connection step. In the edge connection process, starting from all confirmed edge points, the edge points to be determined in their neighborhood are traversed. If the gradient direction of an edge point to be determined is consistent with the gradient direction of a confirmed edge point, it indicates that they belong to the same continuous edge, and the edge point to be determined is upgraded to a confirmed edge point. This process is repeated until all edge points connected to confirmed edge points are verified, ultimately forming continuous and complete edges. The pixels on these edges are the edge feature points of the image.

[0028] For example, in the defect detection of mobile phone OLED screens, taking a bright screen image as an example, fine scratches on the screen surface are directly marked as defined edge points because their grayscale value changes abruptly from 255 to 230, and their Sobel gradient magnitude exceeds the high threshold of 100. Meanwhile, the grayscale value of dark spots gradually changes from 255 to 200, with a gradient magnitude between 50 and 100, making them undefined edge points. Uniform bright areas are discarded because their gradient value is below 50. In the edge connection stage, the algorithm starts from the defined edge points of the scratch: the continuous direction of the scratch keeps the gradient direction of adjacent undefined points consistent, and these points are upgraded to defined edge points, ultimately forming a complete scratch outline. For dark spot areas, although the gradient values ​​of the undefined points at their edges are low, they form a closed continuous distribution around the dark spot and a clear boundary with the defined non-edge points of the surrounding normal screen area, thus being judged as valid defect edges. Isolated noise points formed by fingerprint residue on the screen surface are discarded because they lack continuous edge support. In the final result, scratches appear as continuous thin lines, while dark spots appear as closed irregular edges, representing precisely extracted defect edge feature points.

[0029] In step S102, based on the feature point distribution of the enhanced image, the functional area edge contours of the flexible OLED screen are extracted, and the geometric parameters of the functional area edge contours are determined, including: Based on the feature point distribution in the enhanced image, Hough transform is used to detect straight line and curve segments to obtain a preliminary set of boundary points. If there are small discontinuous regions in the preliminary set of boundary points, then Bézier curve fitting is used to connect the discontinuous points to obtain a continuous edge contour candidate set. The ideal functional area outline in the CAD design drawing is used as a template, and shape matching is performed with the candidate set of continuous edge outlines. The outline with the highest matching degree is the continuous edge outline of the functional area. By performing polynomial fitting on the continuous edge contour of the functional area, the curvature and length are calculated to obtain a set of geometric parameters. Based on the set of geometric parameters, a data format conversion is performed to convert the parameter set into a standardized contour description file, thereby obtaining the geometric parameters of the functional area edge contour.

[0030] It is worth noting that in the field of flexible OLED screen inspection, Hough transform based on the feature point distribution of enhanced images to detect straight and curved segments can effectively extract boundary information of screen defects. Hough transform identifies straight lines or curves by mapping points in image space to parameter space. For example, in defect detection of a curved screen phone, the enhanced image shows tiny cracks at the screen edge. Hough transform can detect the straight segments of these cracks, setting the angular resolution to 1 degree and the distance resolution to 1 pixel to generate a preliminary set of boundary points. These point sets typically contain the main outline of the crack, but may have discontinuous regions due to noise or uneven illumination. The detection method for discontinuous regions involves analyzing the preliminary set of boundary points generated by the Hough transform, calculating the Euclidean distance between adjacent boundary points, and if the distance between two points exceeds a preset threshold, such as 3 pixels (based on the common width of screen cracks), and there are no other boundary points within this distance, it is determined to be a discontinuous region.

[0031] It's worth noting that Bézier curves define smooth curves using control points, making them suitable for handling complex surface defects on screens. For a discontinuous crack, selecting the two endpoints and the middle control point generates a cubic Bézier curve, ensuring a smooth contour that closely matches the actual defect shape. The equation for a cubic Bézier curve is: ,in It is the starting point. , It is a control point. This is the endpoint. For example, the edge points extracted by the Canny algorithm are segmented into three segments by screen reflection: the left vertical line segment (including the point...) (10,50) (10,80), missing middle segment (approximately 15 pixels), right diagonal segment (including dots) (30,100) (40, 90). The starting point is determined first during fitting. (10,80) and the finish line (30,100) Calculate the two intermediate control points again, and set the first control point based on the vertical upward slope of the end of the left line segment. (15,85) is used to continue the upward trend on the left; a second control point is set based on the 45° slope from the beginning of the right-side segment upwards to the left. (25,95), thus echoing the direction on the right in advance. As t changes from 0 to 1, the third-order Bézier curve changes from... Departure, via , Guide, smooth transition to This method fills in the missing middle segment and maintains consistency with the tangent direction of the original edges on both sides. It is particularly suitable for irregular boundaries caused by bending in flexible OLED screens, generating contours that more realistically reflect defect morphology.

[0032] After visual inspection and image processing, multiple edge contours are generated to form a continuous edge contour candidate set, including the raw material edge contour (physical boundary of the mother plate) and the functional area contour (AA area). In order to accurately find the functional area contour, the ideal functional area contour in the CAD design drawing is used as a template and is matched with the continuous edge contour candidate set. The contour with the highest matching degree is the continuous edge contour of the functional area.

[0033] In the process of generating the geometric parameter set, polygon fitting uses the least squares method to fit the contour points and generate a polynomial curve. The core objective of the least squares method in polygon fitting is to find an optimal continuous polynomial curve for a series of discrete contour points, minimizing the overall deviation between this curve and all data points. Specifically, when faced with a complex contour composed of multiple pixels, we first need to determine the order of a polynomial, typically second or third order. Then, we construct a system equation with the polynomial coefficients as unknowns, and use the sum of the squares of the perpendicular distances from all contour points to the polynomial curve as the loss function, calculating a specific set of coefficients that globally minimizes this loss function. For example, 100 discrete pixels of the crack contour are extracted, with coordinates ranging from... [100,600] pixels After removing two noise outliers from the [150, 350] pixel range, 98 valid fitting points are retained. Then, a residual sum of squares function is constructed using a third-order polynomial as the fitting model. For coefficients Find the partial derivatives and set them to 0 to form a system of linear equations. Solve for the coefficients: , , , According to the third-order polynomial The radius of curvature and total profile length were calculated, with a mapping relationship of 200 pixels / mm between pixels and actual physical units (this mapping relationship was determined through pre-calibration experiments). In the calculation results, a crack length of 1000 pixels in the image corresponds to 5 mm in reality; a radius of curvature of 5000 pixels at key points corresponds to 25 mm in reality. These parameters quantify the geometric characteristics of the defect and provide accurate data for subsequent analysis.

[0034] It's worth noting that the data format conversion method directly maps geometric parameters to JSON or XML field values. The description file can be in XML or JSON format, recording information such as the contour's coordinates, curvature, and length. For example, converting the crack's geometric parameters into a JSON file includes fields such as the starting coordinates (100, 200), curvature value of 0.5, and length of 5 mm. This standardized file facilitates system storage and cross-platform analysis, making it suitable for automated inspection systems.

[0035] In step S103, if the geometric parameters of the functional area edge contour deviate from the preset parameter threshold beyond the allowable range, the focal length and illumination intensity parameters of the optical system are adjusted to obtain optimized image quality indicators, including: The geometric parameters of the edge contour of the functional area are detected. If the deviation between the detection result and the preset parameter threshold exceeds the allowable range, the focal length parameter of the optical system is iteratively adjusted to obtain the adjusted focal length value. Based on the adjusted focal length value, the illumination intensity parameters of the optical system are optimized using light intensity equalization to obtain an optimized illumination intensity value; By using the adjusted focal length value and the optimized illumination intensity value, the enhanced image is re-acquired, resulting in an optimized image quality index.

[0036] It is worth noting that the preset parameter thresholds are geometric parameters of the contour point data, such as the length or curvature of the contour. In the contour screening criteria, clear acceptable ranges must be set for key geometric parameters. Specifically, the contour length threshold is set to be no less than 5 mm, and the average curvature threshold is set to be no greater than 0.1 mm. -1 These thresholds are critical criteria for determining whether an extracted contour belongs to a valid target in contour quality assessment. They are determined by analyzing the statistical distribution of historical valid contour datasets. Analysis of historical contour feature vector sets reveals that when the contour length is not less than 5 mm, the contour has sufficient spatial scale for reliable geometric analysis; when the average curvature is not greater than 0.1 mm... -1 This indicates that the overall profile is not excessively curved or sharp, conforming to the morphological characteristics of a typical linear or gently curved component; therefore, the aforementioned threshold is derived. Those skilled in the art will understand that a length threshold of 5 mm and a curvature threshold of 0.1 mm are... -1These are empirically calculated values ​​based on general structural component inspection scenarios. In practical applications, they can be adjusted according to the size and deformation characteristics of the target object. If the geometric parameters of the initial contour point data deviate from the preset parameter threshold by more than 10%, the focal length parameters of the optical system need to be adjusted. The 10% threshold deviation is based on the accuracy requirements of screen contour detection and was determined through extensive sample verification. It was set based on the fact that 95% of the samples in historical data could not meet the accuracy requirements when the deviation exceeded 10%. An iterative algorithm can be used for this adjustment process. The system first extracts the contour, calculates its geometric parameters and the initial deviation from the parameter threshold, and sets the maximum number of iterations to 100. Iteration begins. If the current deviation exceeds 10%, the focal length value is adjusted by 0.1 mm each time, depending on the positive or negative direction of the deviation. The change in the deviation of the contour point data is observed until the deviation converges to within the threshold deviation of 10% or the maximum number of iterations is reached. The adjusted focal length value, for example, 2.5 mm, can more clearly capture the feature points of the functional area edges, thereby improving the accuracy of contour extraction.

[0037] It's worth noting that intensity equalization optimization is a key technology in image processing for improving uneven brightness distribution. Its core principle is to adjust the intensity of light in different areas of the image using algorithms, eliminating localized overexposure or underexposure caused by factors such as light source angle deviations, optical system shadows, or scene reflections. This results in a more uniform overall image brightness while preserving the detailed features of the target area. This optimization process typically begins by analyzing the image's intensity distribution using a grayscale histogram to identify overexposed or underexposed areas. Then, differentiated intensity adjustments are made to different areas: pixel grayscale values ​​are appropriately increased in dark areas to enhance details, while grayscale values ​​are appropriately decreased in bright areas to suppress overexposure. The adjustment range is dynamically adapted based on the intensity correlation of surrounding pixels to avoid color banding. The final output is an image with balanced intensity. For example, for an image with uneven brightness distribution, firstly, a grayscale histogram of the original image is generated. The histogram shows that the peak value in the dark area corresponds to a grayscale value of 40, while the peak value in the bright area corresponds to a grayscale value of 230. Furthermore, the pixel proportion in the middle grayscale range is only 15%, indicating a significant brightness discontinuity. Next, a region-specific adjustment is initiated: for the dark edges, the gain is dynamically increased based on the average grayscale value within a 3×3 pixel block. For example, for a block with an average grayscale value of 40, the gain coefficient is set to 1.8, increasing the grayscale value to 70-80. The gain coefficient is a real number multiplication factor greater than 0 that directly affects the average grayscale value of the image pixel blocks, linearly amplifying or reducing their brightness intensity. Using a peak grayscale value of 40 in the dark area as a baseline, the goal is to boost it to the middle grayscale range of 70-80. The gain coefficient can be calculated by dividing the target grayscale lower limit by the peak grayscale value in the dark area, i.e., 70 ÷ 40 = 1.75. Considering the subtle grayscale differences that may exist in the dark area pixels, to avoid overexposure due to excessive boosting, a certain adjustment redundancy needs to be retained; therefore, the coefficient is rounded to 1.8. The specific boosting method is to multiply the average grayscale value by a pre-calculated gain coefficient to obtain the grayscale boost value. Simultaneously, the grayscale gradient of surrounding blocks is referenced to avoid significant color differences with adjacent areas after adjustment. For the central overexposed area, a grayscale compression strategy is used, proportionally compressing the grayscale values ​​of 220-255 to 160-190, preserving texture details in the overexposed area. The final output balanced image grayscale histogram exhibits a single-peak normal distribution, with the proportion of pixels in the middle grayscale range increasing to 65%. This optimized illumination intensity value reduces shadows or overexposed areas in the image, providing higher quality input for subsequent image processing.

[0038] It's worth noting that the re-acquired enhanced image can be further processed using image quality assessment algorithms. A commonly used assessment method is based on the structural similarity index. The structural similarity index quantifies the degree of image distortion by comparing three dimensions: brightness, contrast, and structure. Its calculation logic is as follows: first, calculate the average grayscale value of the original image... Contrast standard deviation , and the average gray value of the processed image Contrast standard deviation and the covariance of the two Then according to the formula , where constant parameters , , , It is a constant much smaller than 1, and its default value is usually 1. =0.01 and =0.03, It is the dynamic range of the image's pixel values. For example, for an 8-bit grayscale image, =255. The closer the SSIM value is to 1, the more similar the brightness distribution, detail contrast, and texture of the two images, and the less image distortion. If the value is below 0.8, significant distortion, such as blur, noise, or brightness deviation, is generally considered to exist. For example, comparing the original image and the optimized image, an increase in the SSIM value from 0.75 to 0.95 indicates a significant improvement in image sharpness and edge detail. This high-quality image provides more reliable feature point input for the Hough transform, thereby generating more accurate contour point data.

[0039] In step S104, based on the optimized image quality index, the functional area edge feature point sequence is re-extracted to obtain a precise set of functional area edge point coordinates, including: Based on the optimized image quality index, the edge feature point sequence is extracted again from the enhanced image using Hough transform to obtain a preliminary edge point dataset. Contour fitting and matching filtering are then performed to obtain a preliminary edge point dataset for the functional area. If the geometric parameters of the preliminary edge point dataset of the functional area exceed the preset geometric parameter threshold, the edge point coordinates are adjusted to obtain an accurate set of functional area edge point coordinates.

[0040] It's worth noting that edge feature point extraction based on enhanced images is a crucial step in ensuring screen quality. The Hough transform, a classic method, effectively identifies the geometric features of edges by mapping pixels in the image space to a parameter space. Any point in the image space... Corresponding parameter space A sine curve in , The distance from the origin to the line is denoted as . For the normal to the line and Angle between axes; if multiple points are collinear in image space, their sine curves in parameter space will intersect at the same point. The algorithm uses an accumulator to count the number of intersections at each point in the parameter space. When the accumulated value at a point exceeds a preset accumulation threshold, it can be determined that there is a line in the image that intersects with the parameter space. The threshold is a straight line with parameters. The cumulative threshold is typically based on the average density of edge pixels in the image, and is 70% of the number of edge pixels in that region. This value is determined through extensive sample validation based on the accuracy requirements of screen contour detection. If the pixel concentration in the edge region is high, the threshold can be set to 60%-70% of the number of edge pixels in that region to ensure effective extraction of continuous straight lines; if the image has a lot of noise interference, the threshold needs to be appropriately increased to 80% to filter out false straight lines formed by noise points. For example, when detecting a high-resolution flexible OLED screen, the Hough transform can be set with an angular resolution of 0.3 degrees and a distance resolution of 0.6 pixels to generate a preliminary edge point dataset for the functional area. This point data can usually outline the main contours of the functional area edges, but due to possible minor folds or material reflection differences on the screen surface during manufacturing, some edge points may deviate from their expected positions. Based on the preliminary edge point dataset of the functional area, contour fitting is performed using Bézier curve fitting. The Bézier curve fitting method has been explained in step S102 and will not be repeated here. The fitted contour is then matched and filtered using the ideal functional area contour in the CAD design drawing. This matching and filtering method has also been explained in step S102 and will not be repeated here. After filtering, the preliminary edge point dataset of the functional area can be obtained.

[0041] It's worth noting that the preset geometric parameter thresholds are for the geometric parameters of the contour point data, such as the contour's length or curvature. In the contour screening criteria, clear acceptable ranges must be set for key geometric parameters. For example, a length threshold of 5 mm and a curvature threshold of 0.1 mm are used. -1 These are empirically calculated values ​​based on general structural component detection scenarios. In practical applications, adjustments can be made according to the size and deformation characteristics of the target object. If the geometric parameters of the initial edge point dataset of the functional area deviate from the preset parameter threshold by more than 8%, the edge point coordinates need to be optimized through an iterative algorithm. The 8% parameter threshold deviation is based on the accuracy requirements of screen contour detection and was determined through verification with a large number of samples. It was set based on the fact that 95% of the samples in historical data could not meet the accuracy requirements when the deviation exceeded 8%. The core of the iterative algorithm lies in gradually correcting points with large deviations through multiple adjustments. For example, the initial edge points may have coordinate offsets due to uneven lighting. In this case, the coordinate values ​​of the points can be dynamically adjusted by analyzing the spatial distribution of adjacent points, gradually approaching the true edge.

[0042] In one embodiment, each iteration can be based on a weighted average method. The core is to optimize the edge point coordinates by dynamically assigning weights: for points that deviate closely from the local fitted curve, a maximum weight of 1.0 is assigned when the deviation is 0 pixels, completely preserving the original position information of the edge point; within the deviation range of 0-0.25 pixels, the weight is 0.8; within the deviation range of 0.25-0.49 pixels, the weight is 0.6; when the deviation reaches 0.5 pixels, the weight is 0.5, weakening the impact to half to reduce noise interference; within the deviation range of 0.5-0.75 pixels, the weight is 0.2; within the deviation range of 0.75-1.0 pixels, the weight is 0.1, gradually eliminating distant invalid points; when the deviation is ≥1.0 pixels, the weight is fixed at 0, eliminating invalid points and making the adjustment towards the fitted curve more significant. In each iteration, the algorithm first performs local curve fitting on the current edge point, preferentially using a quadratic polynomial curve, calculating the vertical distance from each point to the curve, and then inferring the weight based on the distance. Let the original coordinates of a certain edge point be... Its corresponding point on the quadratic polynomial fitting curve is The perpendicular distance from the point to the curve is Based on distance weights The adjusted coordinates are Final Press , Let be the y-coordinate of the point corresponding to the fitted curve, where the sign function is... Used to determine the adjustment direction, when the edge point is above the curve. , A value of 1 indicates the coordinates are adjusted downwards; when the edge point is below the curve. , A value of -1 indicates that the coordinates are adjusted upwards. The baseline value for each adjustment is the x-coordinate, which is based on x... , Let x be the x-coordinate of the point corresponding to the fitted curve, where the sign function is... Used to determine the adjustment direction; when the edge point is to the right of the curve. , A value of 1 is used to adjust the coordinates to the left, when the edge point is to the left of the curve. , A value of -1 indicates that the coordinates are adjusted to the right. This serves as the baseline value for each adjustment. The points are adjusted towards the curve using the method described above. For example, if the original coordinates of an edge point are (50, 30), and the deviation from the fitted curve is 0.6 pixels with a corresponding weight of 0.3, and the corresponding point on the curve is (50.1, 29.9), then the adjusted coordinates are (50.07, 29.93), approximately 0.2 pixels closer to the curve than the original point. Setting the maximum number of iterations to 100, after multiple iterations, the edge points gradually conform to the true contour curve. When the average deviation of all points converges to within 0.1 pixels, or when the maximum number of iterations is reached, the high-precision edge coordinates after denoising are obtained. This method effectively reduces noise interference and improves the accuracy of edge point coordinates.

[0043] In step S105, the precise set of functional area edge point coordinates is sorted and transformed to generate an initial cutting path curve. The matching degree between the initial cutting path curve and the functional area edge is calculated, and the path offset region in the initial cutting path curve is obtained, including: Based on the set of coordinates of the edge points of the functional area, the point sequence is sorted to obtain an ordered set of point sequences; For the ordered set of points, an affine transformation is used to perform coordinate transformation to obtain the transformed set of points. If the smoothness parameter of the transformed point sequence set deviates from the preset smoothness threshold by more than the allowable range, then the curve trajectory is adjusted by Bézier curve fitting and a certain distance of overall offset is made to generate the initial cutting path curve. Based on the initial cutting path curve, the smoothness and the matching degree of the functional area edge are detected by boundary comparison to obtain the path offset area.

[0044] It's worth noting that the minimum path algorithm can be used to sort the point sequence for the set of functional area edge point coordinates, with the aim of generating an ordered set of point sequences. The core of the minimum path algorithm is to first select a reference point from the unordered set of functional area edge points—usually the point at the very edge of the coordinates—and use it as the first point of the ordered sequence, removing it from the original set. Next, taking the last point of the current sequence as the current point, the Euclidean distance between it and all remaining points in the original set is calculated. Candidate points with a distance less than a preset distance threshold of 0.5 pixels are selected. This 0.5-pixel threshold is determined through extensive sample validation based on the screen detection accuracy requirements. If the threshold is less than 0.5 pixels, it may be overly stringent, causing some truly adjacent edge points to be excluded, resulting in sequence breaks. If the threshold is greater than 0.5 pixels, it may misjudge noise points on the screen surface as adjacent to true edge points, leading to invalid points mixed into the sequence and affecting the accuracy of subsequent contour fitting. If multiple candidate points exist, the point with the smallest distance is selected as the next ordered point, added to the sequence, and removed from the original set. If no point exists with a distance less than a threshold, the threshold is appropriately relaxed, expanding to 0.8 pixels, or the point with the smallest current distance is selected to ensure sequence continuity. For example, setting the distance threshold to 0.5 pixels, the algorithm prioritizes connecting point pairs with a distance less than this threshold, gradually constructing a complete edge trajectory. This method ensures that the point sequence reflects the actual geometry of the screen edge, providing a reliable basis for subsequent transformations.

[0045] It's worth noting that affine transformations, through operations such as translation, rotation, or scaling, map a sequence of points to a standard coordinate system to correct coordinate offsets caused by deviations in the angle or position of the detection device. First, the original coordinates captured from the screen... Mapped to standard coordinate system coordinates Substitute into the affine transformation formula The six transformation parameters were solved. .in Control rotation, scaling, and shearing. The translation is controlled to ultimately form a complete affine transformation matrix. Transformation parameters can be used for translation, rotation, or scaling operations. For example, when inspecting a flexible OLED screen, a slight tilt in the camera angle may distort the edge point sequence. An affine transformation can rotate the point sequence by 5 degrees and translate it by 0.3 pixels, correcting it to a standard plane. This transformation unifies the coordinate reference of the point sequence, facilitating subsequent analysis.

[0046] For the ordered set of points, an affine transformation is used to perform coordinate transformation to obtain a transformed set of points; this set of points can then generate a curve trajectory. If the smoothness parameter of the transformed set of points deviates from the preset smoothness threshold by more than the allowable range, a Bézier curve fitting is used to adjust the curve trajectory.

[0047] Subsequently, by offsetting the curve trajectory by a certain distance (which is reasonably selected according to actual processing needs), the initial cutting path curve can be generated. The initial cutting path formed in this way has the following advantages: protecting the core functional area. The light-emitting area (Active Area, or AA area) of the flexible OLED screen is an extremely fragile and critical part. The cutting path (i.e., the center of the laser beam) cannot directly pass through the AA area, otherwise it will directly damage the screen; reserving a safety margin. Cutting will generate a heat-affected zone, micro-cracks, and stress. Reserving a safety margin around the AA area can absorb these negative effects and ensure the integrity and performance of the AA area; compensating for alignment errors. No matter how precise the vision system is, there are always small alignment errors. This offset is to accommodate these errors and ensure that nothing goes wrong.

[0048] It's worth noting that smoothness parameters are typically evaluated based on the curvature changes of a point sequence. If the curvature deviation exceeds 10%, it indicates abrupt changes or noise points in the edge trajectory. Deviations exceeding 10% are derived from historical experience data. In one possible implementation, a smooth curve can be generated by fitting a cubic Bézier curve to the edge points of the functional area of ​​a flexible OLED screen. For the marked non-smooth areas, smoothing optimization is performed using a cubic Bézier curve. The key lies in the selection and iterative adjustment of control points: First, four key control points are selected from the edge point sequence. The first and last points are the start and end points of the non-smooth area, while the two middle control points are initially set based on the direction of the surrounding points. For example, when fitting an arc-shaped edge of the screen, the fixed endpoints at both ends of the edge are first determined, and then the curvature trend of the edge between the two endpoints is observed. If the edge arches slowly upward from left to right, two intermediate points are roughly divided between the two endpoints along the arched curvature contour so that they can connect the entire curved contour of the edge. Then, the coordinates of the control points are optimized by weighted adjustment, with the weight allocation based on the principle of being close to the actual edge points. Specifically... First, based on the pixel-level detection accuracy requirements of OLED screens (typically the error needs to be ≤1 pixel), the distance from the initially set control points to the actual edge points is divided into three core intervals: 0-0.2 pixels is the near interval, assigned a high weight of 0.7, because the actual edge points in this interval are closest to where the control points should be; 0.2-0.35 pixels is the middle interval, assigned a medium weight of 0.3, as these edge points can help correct control point deviations, but their influence needs to be weaker than the near interval; 0.35-0.5 pixels is the far interval, assigned only a low weight of 0.1 to avoid excessively far edge points interfering with the control points' fit to the actual edge; and edge points with a distance exceeding 0.5 pixels are directly assigned a weight of 0, as they are outside the effective reference range. After calculating the theoretical optimization value based on the weighted average, the optimization value is gradually approached by a single adjustment of 0.1 pixels to avoid over-correction leading to trajectory distortion, thus gradually bringing the curve closer to the actual edge. For example, the starting and ending points are at coordinates (20,50) and (50,60), and the two control points in the middle are at (30,52) and (40,58). After the first adjustment, the middle control points become (30.1,52) and (39.9,58), and the average distance between the curve and the edge points decreases from 0.3 pixels to 0.25 pixels. After repeating the adjustment 5-8 times, the distance between the curve and all edge points is controlled within 0.15 pixels. The areas that originally had abrupt changes become smooth, and the natural curvature of the flexible OLED screen is completely preserved, providing stable trajectory data for subsequent contour parameter calculations.

[0049] It's worth noting that the boundary comparison algorithm identifies path offset regions by comparing the pixel differences between the fitted curve and the functional area edges. In its implementation, the algorithm first acquires two key data points: one is a standard edge curve generated through Bézier curve or polynomial fitting, and the other is the actual set of edge pixels obtained from image acquisition and functional area edge extraction of the screen to be detected. Then, following the same... coordinates or The coordinate intervals are used to calculate the Euclidean distance between each point on the standard curve and its corresponding pixel on the actual edge. A pixel deviation threshold of 0.4 pixels is then set. This value, based on the accuracy requirements of screen inspection, has been verified through a large number of samples. In actual inspection, if three consecutive pixels are detected with deviations exceeding this threshold, or if any single pixel's deviation exceeds twice the threshold, the segment is marked as a path offset region. For example, in flexible OLED screen inspection, the point at (150, 200) on the standard curve has a deviation of 0.5 pixels from its corresponding pixel at (150, 200.5) on the actual edge, exceeding the 0.4 pixel threshold. Furthermore, its adjacent points (151, 200.2) and (152, 200.6) also have deviations of 0.42 and 0.58 pixels respectively. Therefore, the region with x-coordinates 150-152 is marked as a path offset region. By recording the coordinate range of these regions, accurate data support can be provided for subsequent quality assessment.

[0050] In step S106, the smoothness of the initial cutting path curve is calculated. If the smoothness is lower than a preset smoothness threshold, the path offset region is adjusted. If the updated state of the path offset region of the adjusted cutting path curve meets a preset standard, a qualified cutting path curve is obtained, including: Obtain the adjusted cutting path curve, extract curvature distribution features, and obtain the curvature distribution set; If there are points in the curvature distribution set whose curvature values ​​exceed a preset threshold, then vector correction is used to adjust the coordinates of the corresponding points to obtain a set of correction vectors; Using the set of correction vectors, the point sequence set is updated by coordinate transformation to obtain the updated point sequence set; For the updated set of point sequences, edge detection is used to calculate the matching degree with the edge of the functional area, and the path offset region update status of the adjusted cutting path curve is determined. If the path offset region update status of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained.

[0051] It's worth noting that for the adjusted cutting path curve, the curvature calculation method analyzes the curvature value at each point along the path, forming a curvature distribution set. When inspecting the edge of a functional area of ​​a flexible OLED screen, assuming a path contains 500 points, curvature calculation reveals that the curvature value at a certain point reaches 0.25, while the preset curvature threshold is 0.15. This curvature threshold, determined through extensive sample verification based on the screen inspection accuracy requirements, indicates that this point may have experienced excessive local bending due to minor defects in the manufacturing process. This method can accurately locate points with abnormal curvature, providing a basis for subsequent corrections.

[0052] It's worth noting that the vector correction algorithm is used to adjust the coordinates of points whose curvature values ​​exceed a threshold, generating a set of correction vectors. The vector correction algorithm analyzes the local geometric features of a point to calculate the adjustment direction and magnitude, thus smoothing curvature changes. Specifically, if the curvature value of a point exceeds the threshold, the vector correction algorithm can calculate its relative displacement with neighboring points to generate a correction vector. First, the coordinates of the point P with curvature exceeding the threshold and its left and right adjacent normal points P1 and P2 are obtained, and vectors P1P and PP2 are calculated. The angle or cross product of the two vectors determines the current curvature offset direction. Then, based on the specific value of the curvature exceeding the threshold, such as a deviation threshold of 0.1 pixels (this deviation threshold of 0.1 pixels is determined through extensive sample verification based on the screen detection accuracy requirements), the required displacement length for that point is calculated. Finally, using the normal direction of the line connecting adjacent points P1P2 as a reference, the displacement length and direction are combined to generate a set of correction vectors that can adjust the curvature value of the point. The correction vector is pulled back to a reasonable curvature range to ensure that the corrected point sequence still conforms to the overall contour of the material edge. For example, assuming that the coordinates of a point need to be translated outward by 0.2 pixels to reduce curvature, the algorithm will generate a corresponding correction vector, recording the adjustment direction and distance. This method can effectively smooth local anomalies and ensure the continuity of the path. For example, the deviation curvature of point P (50,30) is 0.2 pixels, which exceeds the deviation threshold of 0.1 pixels. The left adjacent normal point P1 (48,28) and the right adjacent point P2 (52,32) are calculated. The vectors P1P and PP2 are (50-48,30-28) and (52-50,32-30), which are (2,2) and (2,2), respectively. Obviously, the cross product of P1P and PP2 is 0 (indicating that P is on the path, without offset, but the curvature exceeds the limit). Through: Calculate the displacement length that needs to be adjusted, where The scaling factor ensures that the displacement is proportional to the curvature deviation, avoiding over-adjustment while reaching the deviation correction point. Its value is determined through calibration based on historical data. For example, It is 0.5. The deviation curvature allows us to calculate the required adjustment length of the displacement. The value is 0.5 × 0.2 = 0.1 pixels. Then, for P1P2 (52-48, 32-28), i.e., (4,4), the normal vector is determined as (-4,4). The normal vector is then normalized using its magnitude, which is approximately 5.66. After normalization, we obtain (-4 / 5.66, 4 / 5.66), i.e., (-0.707, 0.707), thus obtaining the corrected vector direction. Multiplying these values ​​yields the correction vector (-0.071, 0.071) (rounded to three decimal places). Then, P is corrected using this correction vector, resulting in the corrected coordinates (50-0.071, 30+0.071), or (49.929, 30.071).

[0053] It's worth noting that the coordinate transformation method, based on the set of correction vectors, is used to update the point sequence set. The coordinate transformation applies the correction vectors to the point sequence through global or local adjustments, generating an updated set of point sequences. For example, when detecting a flexible OLED screen, if a path segment is offset by 0.3 pixels due to device calibration errors, the coordinate transformation method can adjust the entire point sequence to the correct position through a translation operation. For instance, the coordinates of a point sequence can be translated 0.2 pixels to the right while rotating 2 degrees to align with the standard edge. This method unifies the coordinate system of the path and the actual edge, improving the accuracy of subsequent matching.

[0054] It's worth noting that, for the updated set of point sequences, the edge detection algorithm calculates the matching degree with the functional area edges to determine the update status of path offset regions. The matching degree is determined by comparing the pixel differences between the point sequence and the actual functional area edges, and it needs to consider a certain distance of offset from the curve trajectory mentioned above. The matching threshold is determined based on the screen detection accuracy requirements and the offset distance, and is verified through a large number of samples. Specifically, when detecting flexible OLED screens, if the pixel deviation of a certain path segment is lower than the preset matching threshold, it indicates that the path segment highly matches the edge; if another segment has a deviation higher than the matching threshold, it is marked as a path offset region, requiring further optimization. For example, the edge detection algorithm can generate a matching degree distribution map by comparing point by point, intuitively displaying the update status of the path offset region. This method can provide accurate data support for quality control.

[0055] The process of obtaining the adjusted cutting path curve includes: Based on the initial cutting path curve, the path smoothness parameter is calculated using the least squares method to obtain the set of deviation point coordinates; For the set of deviation point coordinates, gradient descent iteration is used to identify deviation points and obtain a set of correction point weights; Based on the set of correction point weights, an affine transformation is used to adjust the coordinates of the deviation points to obtain a set of adjusted point sequences. For the adjusted point sequence set, the matching degree between the boundary comparison detection and the functional area edge is obtained to obtain the adjusted cutting path curve.

[0056] It's worth noting that the least squares method generates a smoothness parameter by fitting the deviation between the point sequence and the ideal smooth curve. For example, when inspecting a high-resolution flexible OLED screen, the initial path curve may contain thousands of edge points, some of which may have uneven trajectories due to minor deformations during production. If the sum of squares of the deviations for a certain path segment is 0.3 pixels, it can be determined that the smoothness is insufficient and requires further processing. This method can effectively identify irregular points, providing a basis for subsequent optimization.

[0057] It is worth noting that, for the set of deviation point coordinates, a gradient descent algorithm is used to iteratively identify the deviation points and generate a set of correction point weights. The gradient descent algorithm iteratively optimizes the objective function as the weighted sum of the squared distances from the deviation points to the baseline path and the rate of curvature change of adjacent points. Minimize the impact of point position weights by gradually adjusting them to filter out points that have a significant influence on path smoothness. Using the set of deviation point coordinates as initial input, construct an objective function that is the weighted sum of the squared distances from deviation points to the baseline path and the rate of curvature change of adjacent points. The minimum number of iterations is set, and the iteration termination threshold is set to a weight adjustment amount less than 0.001. =100; In each iteration, calculate the gradient of the objective function with respect to the weights of each deviation point. This gives us the direction of weight adjustment. Then, based on the preset learning rate =0.01, calculate the weight update amount, repeat the iteration process until the objective function value converges or the maximum number of iterations is reached, and finally generate the set of corrected point weights.

[0058] For example, when detecting the edges of functional areas on a flexible OLED screen, if the curvature change at a point exceeds 0.2 pixels, it can be assigned a higher weight using a gradient descent algorithm and marked as a deviation point. The number of iterations can be set to 50, with the weight adjusted by approximately 0.01 each time, ensuring recognition accuracy. This method can accurately locate points that need correction, providing reliable input for subsequent transformations.

[0059] It is worth noting that, based on the set of corrected point weights, an affine transformation method is used to adjust the coordinates of the deviation points, generating an adjusted point sequence set. The affine transformation corrects the positional deviations of the deviation points through translation or rotation operations. The specific rules of the affine transformation follow those determined in step S105. For example, if the detection device causes the edge points to shift by 0.4 pixels due to angular deviation, the point sequence can be translated by 0.2 pixels and rotated by 3 degrees through an affine transformation, mapping it to the standard coordinate system. This transformation unifies the coordinate reference, facilitating subsequent matching analysis. The adjusted point sequence set is closer to the true edge trajectory, providing support for path optimization.

[0060] It is worth noting that, for the adjusted point sequence set, a boundary comparison algorithm is used to detect the matching degree with the functional area edge. The matching degree is set by considering the distance of the curve trajectory offset mentioned above, and the adjusted cutting path curve is determined. The boundary comparison algorithm quantifies the matching degree by comparing the pixel difference between the fitted curve and the actual edge.

[0061] In step S107, a cutting operation is performed according to the qualified cutting path curve to obtain a finished flexible OLED screen, including: The qualified cutting path curve is smoothed by path optimization, and the point coordinates are adjusted by interpolation calculation to obtain a smoothed point sequence set. Based on the smoothed point sequence set, a control command sequence recognizable by the laser device is generated using path mapping. The deviation is corrected by coordinate transformation to obtain the corrected control command set. For the corrected set of control commands, the dynamic parameters of the laser equipment during operation are monitored in real time. If the dynamic parameters deviate from the preset threshold, the command sequence is adjusted and updated through feedback to obtain the updated command sequence. The updated instruction sequence is transmitted to a preset laser device via a device driver interface, which then controls the preset laser device to perform a cutting operation, resulting in a finished flexible OLED screen.

[0062] It's worth noting that in the field of flexible OLED screen cutting, the core of path optimization algorithms lies in adjusting point coordinates through interpolation calculations to generate a smooth set of point sequences. Specifically, interpolation calculations can employ spline interpolation methods, generating a smooth transition curve by analyzing the coordinate differences between adjacent points on the path. (Cubic spline interpolation function) ,in , , and Let be the polynomial coefficients of this subinterval. For example, the algorithm first preprocesses the 300 points of the original cutting path to identify the resulting jagged fluctuation regions. This is typically done by calculating the slope changes of adjacent points; if the slope fluctuation of five consecutive points exceeds 0.2, it is determined to be a region requiring optimization. Then, a cubic spline interpolation method is used: taking the beginning and end points of the fluctuation region as boundary conditions, points A(98,199) and B(102,201), and selecting the two middle fluctuation points (99,200.5), (100,199.2), and (101,200.8) as known control points, a cubic spline function is constructed. The function must satisfy the conditions of continuous function values, continuous first derivative, and continuous second derivative at adjacent control points to ensure that the generated curve has no sharp corners. Through interpolation calculations, two new points (99.3,200.1) and (99.7,199.5) are inserted between points (99,200.5) and (100,199.2), while the original fluctuating point (100,199.2) is adjusted to (100.2,199.8), so that the slope of this path smoothly transitions from 0.3 to -0.1, and the rate of change of curvature decreases from 0.15 to below 0.05. The optimized 300 points form a smooth point sequence. When moving along this trajectory during laser cutting, rough cutting edges caused by jagged paths can be avoided.

[0063] It's worth noting that, based on the smoothed point sequence set, the path mapping method aims to transform the point sequence into a control command sequence recognizable by the laser device. Specifically, path mapping generates a command sequence containing speed, direction, and power by converting point coordinates into motion trajectory parameters driven by the device. For example, assuming a path needs to be cut at a speed of 2 mm / s, path mapping converts the coordinate differences of the point sequence into time steps and laser power values, such as mapping a point (150, 250) to the command "move to (151, 250.2) at 1.5 mm / s, power 80%". This method ensures compatibility between the commands and the device hardware, improving the execution efficiency of the cutting operation.

[0064] In one implementation, a coordinate transformation algorithm is used to correct deviations in the command sequence to ensure that the cutting path is aligned with the actual edge. Specifically, the algorithm adjusts the target points in the command sequence by analyzing the coordinate offset during device operation. For example, if the laser device experiences a systematic offset of 0.1 mm due to thermal expansion, the coordinate transformation algorithm will shift all command points to the left by 0.1 mm, such as correcting command point (200, 300) to (199.9, 300). This method effectively eliminates the influence of device errors and ensures the accuracy of the cutting path.

[0065] It's worth noting that the real-time monitoring algorithm ensures the stability of the cutting process by detecting dynamic parameters during laser equipment operation. Specifically, these dynamic parameters include laser power, cutting speed, and equipment vibration frequency. For example, if the laser power is detected to drop to 70% at a certain moment, below the preset laser power threshold of 80% (this 80% threshold is determined through extensive sample verification based on screen detection accuracy requirements), the real-time monitoring algorithm will record this anomaly and trigger a feedback mechanism. This method can promptly detect deviations during operation, providing a basis for subsequent adjustments.

[0066] In one implementation, the feedback adjustment algorithm dynamically updates the command sequence to address parameter deviations by analyzing real-time monitoring data. Specifically, if the cutting speed is detected to have slowed to 1.8 mm / s due to changes in material hardness, the feedback adjustment algorithm recalculates the command sequence and adjusts it to 2.2 mm / s to maintain efficiency. First, if the cutting speed is detected to have slowed to 1.8 mm / s due to changes in material hardness, the feedback adjustment algorithm recalculates the command sequence and calculates a deviation of 0.2 mm / s from the preset target parameter, such as the standard cutting speed of 2.0 mm / s. Then, the algorithm calls a built-in parameter correlation model, which uses the hardness change value, the initial cutting speed, and the deviation magnitude as model input variables. Linear regression is used to establish the mapping relationship between the input and output. The specific expression is: ,in For speed compensation value, This represents the increase in hardness. The initial velocity, The deviation amplitude is represented by 0.005, -0.1, and 1.2, which are characteristic coefficients, and 0.02 is a constant term. The characteristic coefficients and constant term were calculated from 5000 historical tagged samples trained using the least squares method. These coefficients were confirmed to minimize the average error between the predicted compensation value and the actual value, and were thus determined as the final coefficients. The speed compensation value is the model's output variable. For example, when the hardness increases by 40 HV, the initial speed is 2.0 mm / s, and the deviation amplitude is 0.2 mm / s, the calculated compensation value is 0.2 mm / s, meaning the speed needs to be increased from 2.0 mm / s to 2.2 mm / s. Finally, the algorithm converts the calculated new parameter of 2.2 mm / s into a command signal recognizable by the equipment, overwriting the corresponding parameters in the original command sequence, thus completing the dynamic update. For example, the command sequence for a certain path segment is updated to "move to (160, 260) at 2.2 mm / s" to adapt to the material properties. This method can dynamically optimize the cutting process and improve the consistency of the finished product.

[0067] It's worth noting that the device driver interface controls the laser device to perform the cutting operation by transmitting an updated sequence of instructions. Specifically, the interface must ensure that the instruction sequence is sent in the correct format and timing. For example, an instruction sequence transmitted in G-code format, such as "G01X150Y250F2.2", indicates a movement to coordinates (150, 250) at 2.2 mm / s. This method ensures efficient instruction execution, ultimately producing a standard-compliant flexible OLED screen.

[0068] In step S108, based on the finished flexible OLED screen, the edge quality of the finished screen is evaluated, and the electrical and optical performance of the finished screen is tested. Through parameter adjustment and real-time feedback adjustment, a stable matching final path is obtained, including: Based on the finished flexible OLED screen, the edge contour curve of the finished screen is extracted by image segmentation to evaluate the edge quality of the finished screen. Optical and electrical tests were conducted on the finished screen to obtain the corresponding test results. Based on the edge quality and test results, the laser cutting path is adjusted according to feedback to obtain a stable and matched final path.

[0069] Specifically, edge quality data can include smoothness curves and defect distribution data (the number, size, and location of microcracks, chipped edges, burrs, etc.), outputting a quantitative edge quality report containing numerical values ​​for each indicator, a visual deviation map, and a defect coordinate map. Electrical and optical performance testing is used to verify the screen's functionality and rule out potential internal damage and performance inconsistencies. Examples include open / short circuit testing: checking whether the screen edge cutting area is damaged by laser heat or mechanical stress; resistance uniformity testing: measuring the sheet resistance of ITO or other transparent electrodes to ensure uniform conductivity across the entire screen surface, without resistance changes caused by stress concentration introduced during the cutting process; brightness and color uniformity: focusing on brightness and color attenuation in the edge areas. An electrical and optical performance test report can be output, including all test parameters, performance distribution maps, and location information of defective pixels / functional defects.

[0070] By utilizing machine learning (such as neural networks and reinforcement learning), edge quality indicators and performance test results are used as inputs, and cutting path adjustment parameters are used as outputs to train a predictive model. This model then derives the optimal combination of cutting path adjustment parameters, resulting in a stable and matching final path. For example, if edge detection detects an inward shift in the upper left corner of the path, and electrical testing finds a line defect at the same location, path adjustment (geometric compensation) is performed: global coordinate compensation is applied to the entire cutting path. Based on the measured -8μm deviation at the upper left corner, combined with deviation data from other areas, the system calculates an average deviation of -3μm (overall inward bias). Therefore, the entire cutting path is offset outward by +5μm to ensure it never cuts into the display area, while also considering the edge encapsulation of subsequent processes. Ultimately, through targeted adjustments, a dynamic and non-uniform "optimal cutting path" is generated, transforming laser cutting from a "static" process into an "adaptive" intelligent manufacturing process.

[0071] It's worth noting that image segmentation algorithms are used to extract the set of pixels for edge contours. The core of this approach lies in identifying continuous pixels at the edges of the finished screen by analyzing differences in image grayscale values. Specifically, edge detection-based segmentation methods can be employed. Gaussian filtering smooths image noise, and the Canny algorithm is then used to extract edge pixels. For example, assuming a flexible OLED screen image has a resolution of 1920x1080, edge detection yields a contour set containing 5000 pixels. These pixel coordinates, such as (300, 400) and (301, 401), form the initial contour curve. This method ensures the accuracy of contour extraction, providing a reliable data foundation for subsequent processing.

[0072] In summary, this invention acquires the original image of a flexible OLED screen, removes reflections and noise to obtain an enhanced image, and extracts the edge contours and geometric parameters of the functional areas of the flexible OLED screen. When parameters exceed limits, the focal length and illumination intensity of the optical system can be adjusted accordingly. Through multi-level enhancement processing of the original image, the adverse effects on edge recognition accuracy are effectively eliminated. This not only ensures high-precision positioning of the extracted functional area edge point coordinate set but also provides a reliable data foundation for subsequent laser cutting path planning. This invention re-extracts the functional area edge feature point sequence from the newly acquired enhanced image to generate an initial cutting path curve, detects the path curve matching degree and smoothness, and corrects path offset regions. If the updated state of the path offset region of the adjusted cutting path curve meets the standard, it is considered qualified. The cutting path curve is used to perform cutting operations based on the qualified cutting path curve to obtain a finished flexible OLED screen. Through multiple rounds of testing and correction, it is ensured that the final output qualified cutting path curve matches the edge contour of the functional area and has good smoothness. It can be directly used for actual cutting operations, effectively improving cutting efficiency and finished product qualification rate. This invention evaluates the edge quality of the finished screen and conducts electrical and optical performance tests on the finished screen. After adjustment and feedback, a stable matching final path is obtained. Through real-time monitoring and feedback adjustment, the path is dynamically optimized to adapt to variables in production, resulting in a stable path set. The stable path set can be used as a standardized template for cutting similar products, eliminating the need for image processing and path generation from scratch each time. This significantly improves the reusability of the process and cutting efficiency, and reduces time loss in mass production.

[0073] Reference Figure 2 The second embodiment of the present invention provides a schematic diagram of a control system structure for a laser cutting equipment for flexible OLED screens, including: The image acquisition module is used to acquire the original image data of the flexible OLED screen, remove light reflection and noise signals from the original image data, and obtain an enhanced image; The image contour feature extraction module is used to extract the edge contour of the functional area of ​​the flexible OLED screen based on the feature point distribution of the enhanced image, and determine the geometric parameters of the edge contour of the functional area. The image optimization module is used to adjust the focal length and illumination intensity parameters of the optical system to obtain optimized image quality indicators if the geometric parameters of the edge contour of the functional area deviate from the preset parameter threshold by more than the allowable range. The precise functional area contour feature extraction module is used to re-extract the functional area edge feature point sequence based on the optimized image quality index, so as to obtain a precise set of functional area edge point coordinates; The initial cutting path generation module is used to sort and transform the coordinates of the precise functional area edge points, generate an initial cutting path curve, calculate the matching degree between the initial cutting path curve and the functional area edge, and obtain the path offset area in the initial cutting path curve. A qualified cutting path generation module is used to calculate the smoothness of the initial cutting path curve. If the smoothness is lower than a preset smoothness threshold, the path offset region is adjusted. If the update status of the path offset region of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained. The screen cutting module is used to perform a cutting operation according to the qualified cutting path curve to obtain a finished flexible OLED screen. The final cutting path generation module is used to evaluate the edge quality of the finished flexible OLED screen and conduct electrical and optical performance tests on the finished screen. Through parameter adjustment and real-time feedback adjustment, a stable and matched final path is obtained.

[0074] In summary, this invention acquires the original image of a flexible OLED screen, removes reflections and noise to obtain an enhanced image, and extracts the edge contours and geometric parameters of the functional areas of the flexible OLED screen. When parameters exceed limits, optical parameters are adjusted, and the sequence of feature points on the edge of the functional areas is re-extracted to generate an initial cutting path curve. The matching degree and smoothness of the path curve are detected, and the offset area is corrected. If the update status of the offset area of ​​the adjusted cutting path curve meets the standard, a qualified cutting path curve is obtained. Cutting operations are performed according to the qualified cutting path curve to obtain the finished flexible OLED screen. The edge quality of the finished screen is evaluated, and the electrical and optical performance of the finished screen are tested. After adjustment and feedback, a stable and matching final path set is obtained. This method solves the technical problem of insufficient cutting precision, realizes dynamic control from image quality optimization to precise generation of cutting paths, breaks the limitation of independent processing of a single link, and constructs a cutting control system that can correct deviations in real time. This processing method not only significantly improves the fit between the cutting path and the actual functional areas of the screen, but also lays a solid process foundation for the high-precision processing of flexible OLED screens.

[0075] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A control method for laser cutting equipment for flexible OLED screens, characterized in that, include: The original image data of the flexible OLED screen is acquired, and light reflection and noise signals are removed from the original image data to obtain an enhanced image; Based on the feature point distribution of the enhanced image, the edge contours of the functional areas of the flexible OLED screen are extracted, and the geometric parameters of the edge contours of the functional areas are determined. If the geometric parameters of the functional area edge contour deviate from the preset parameter threshold by more than the allowable range, the focal length and illumination intensity parameters of the optical system are adjusted to obtain optimized image quality indicators. Based on the optimized image quality index, the functional area edge feature point sequence is re-extracted to obtain an accurate set of functional area edge point coordinates. The precise set of functional area edge point coordinates is sorted and transformed to generate an initial cutting path curve. The matching degree between the initial cutting path curve and the functional area edge is calculated, and the path offset area in the initial cutting path curve is obtained. Calculate the smoothness of the initial cutting path curve. If the smoothness is lower than a preset smoothness threshold, adjust the path offset region. If the update status of the path offset region of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained. Perform the cutting operation according to the qualified cutting path curve to obtain the finished flexible OLED screen; Based on the finished flexible OLED screen, the edge quality of the finished screen is evaluated, and the electrical and optical performance of the finished screen is tested. Through parameter adjustment and real-time feedback adjustment, a stable matching final path is obtained.

2. The control method for laser cutting equipment for flexible OLED screens according to claim 1, characterized in that, The step of extracting the functional area edge contours of the flexible OLED screen based on the feature point distribution of the enhanced image and determining the geometric parameters of the functional area edge contours includes: Based on the feature point distribution in the enhanced image, Hough transform is used to detect straight line and curve segments to obtain a preliminary set of boundary points. If there are small discontinuous regions in the preliminary set of boundary points, then Bézier curve fitting is used to connect the discontinuous points to obtain a continuous edge contour candidate set. The ideal functional area outline in the CAD design drawing is used as a template, and shape matching is performed with the candidate set of continuous edge outlines. The outline with the highest matching degree is the continuous edge outline of the functional area. By performing polynomial fitting on the continuous edge contour of the functional area, the curvature and length are calculated to obtain a set of geometric parameters. Based on the set of geometric parameters, a data format conversion is performed to convert the parameter set into a standardized contour description file, thereby obtaining the geometric parameters of the functional area edge contour.

3. The control method for laser cutting equipment for flexible OLED screens according to claim 1, characterized in that, If the geometric parameters of the functional area edge contour deviate from the preset parameter threshold beyond the allowable range, the focal length and illumination intensity parameters of the optical system are adjusted to obtain optimized image quality indicators, including: The geometric parameters of the edge contour of the functional area are detected. If the deviation between the detection result and the preset parameter threshold exceeds the allowable range, the focal length parameter of the optical system is iteratively adjusted to obtain the adjusted focal length value. Based on the adjusted focal length value, the illumination intensity parameters of the optical system are optimized using light intensity equalization to obtain an optimized illumination intensity value; By using the adjusted focal length value and the optimized illumination intensity value, the enhanced image is re-acquired, resulting in an optimized image quality index.

4. The control method for laser cutting equipment for flexible OLED screens according to claim 1, characterized in that, The step of re-extracting the functional area edge feature point sequence based on the optimized image quality index to obtain a precise set of functional area edge point coordinates includes: Based on the optimized image quality index, the edge feature point sequence is extracted again from the enhanced image using Hough transform to obtain a preliminary edge point dataset. Contour fitting and matching filtering are then performed to obtain a preliminary edge point dataset for the functional area. If the geometric parameters of the preliminary edge point dataset of the functional area exceed the preset geometric parameter threshold, the edge point coordinates are adjusted to obtain an accurate set of functional area edge point coordinates.

5. The control method for laser cutting equipment for flexible OLED screens according to claim 1, characterized in that, The process of sorting and transforming the precise set of functional area edge point coordinates to generate an initial cutting path curve, calculating the matching degree between the initial cutting path curve and the functional area edge, and obtaining the path offset region in the initial cutting path curve includes: Based on the set of coordinates of the edge points of the functional area, the point sequence is sorted to obtain an ordered set of point sequences; For the ordered set of points, an affine transformation is used to perform coordinate transformation to obtain the transformed set of points. If the smoothness parameter of the transformed point sequence set deviates from the preset smoothness threshold by more than the allowable range, then the curve trajectory is adjusted by Bézier curve fitting and a certain distance of overall offset is made to generate the initial cutting path curve. Based on the initial cutting path curve, the smoothness and the matching degree of the functional area edge are detected by boundary comparison to obtain the path offset area.

6. The control method for laser cutting equipment for flexible OLED screens according to claim 1, characterized in that, The process involves calculating the smoothness of the initial cutting path curve. If the smoothness is lower than a preset smoothness threshold, the path offset region is adjusted. If the updated state of the path offset region of the adjusted cutting path curve meets a preset standard, a qualified cutting path curve is obtained. This includes: Obtain the adjusted cutting path curve, extract curvature distribution features, and obtain the curvature distribution set; If there are points in the curvature distribution set whose curvature values ​​exceed a preset threshold, then vector correction is used to adjust the coordinates of the corresponding points to obtain a set of correction vectors; Using the set of correction vectors, the point sequence set is updated by coordinate transformation to obtain the updated point sequence set; For the updated set of point sequences, edge detection is used to calculate the matching degree with the edge of the functional area, and the path offset region update status of the adjusted cutting path curve is determined. If the path offset region update status of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained.

7. The control method for laser cutting equipment for flexible OLED screens according to claim 6, characterized in that, The process of obtaining the adjusted cutting path curve includes: Based on the initial cutting path curve, the path smoothness parameter is calculated using the least squares method to obtain the set of deviation point coordinates; For the set of deviation point coordinates, gradient descent iteration is used to identify deviation points and obtain a set of correction point weights; Based on the set of correction point weights, an affine transformation is used to adjust the coordinates of the deviation points to obtain a set of adjusted point sequences. For the adjusted point sequence set, the matching degree between the boundary comparison detection and the functional area edge is obtained to obtain the adjusted cutting path curve.

8. The control method for laser cutting equipment for flexible OLED screens according to claim 1, characterized in that, The step of performing a cutting operation according to the qualified cutting path curve to obtain a finished flexible OLED screen includes: The qualified cutting path curve is smoothed by path optimization, and the point coordinates are adjusted by interpolation calculation to obtain a smoothed point sequence set. Based on the smoothed point sequence set, a control command sequence recognizable by the laser device is generated using path mapping. The deviation is corrected by coordinate transformation to obtain the corrected control command set. For the corrected set of control commands, the dynamic parameters of the laser equipment during operation are monitored in real time. If the dynamic parameters deviate from the preset threshold, the command sequence is adjusted and updated through feedback to obtain the updated command sequence. The updated instruction sequence is transmitted to a preset laser device via a device driver interface, which then controls the preset laser device to perform a cutting operation, resulting in a finished flexible OLED screen.

9. The control method for laser cutting equipment for flexible OLED screens according to claim 1, characterized in that, The process of evaluating the edge quality of the finished flexible OLED screen and conducting electrical and optical performance tests on the finished screen, and obtaining a stable matching final path through parameter adjustment and real-time feedback adjustment, includes: Based on the finished flexible OLED screen, the edge contour curve of the finished screen is extracted by image segmentation to evaluate the edge quality of the finished screen. Optical and electrical tests were conducted on the finished screen to obtain the corresponding test results. Based on the edge quality and test results, the laser cutting path is adjusted according to feedback to obtain a stable and matched final path.

10. A control system for a laser cutting equipment for flexible OLED screens, characterized in that, include: The image acquisition module is used to acquire the original image data of the flexible OLED screen, remove light reflection and noise signals from the original image data, and obtain an enhanced image; The image contour feature extraction module is used to extract the edge contour of the functional area of ​​the flexible OLED screen based on the feature point distribution of the enhanced image, and determine the geometric parameters of the edge contour of the functional area. The image optimization module is used to adjust the focal length and illumination intensity parameters of the optical system to obtain optimized image quality indicators if the geometric parameters of the edge contour of the functional area deviate from the preset parameter threshold by more than the allowable range. The precise contour feature extraction module is used to re-extract the functional area edge feature point sequence based on the optimized image quality index, so as to obtain a precise set of functional area edge point coordinates; The initial cutting path generation module is used to sort and transform the coordinates of the precise functional area edge points, generate an initial cutting path curve, calculate the matching degree between the initial cutting path curve and the functional area edge, and obtain the path offset area in the initial cutting path curve. A qualified cutting path generation module is used to calculate the smoothness of the initial cutting path curve. If the smoothness is lower than a preset smoothness threshold, the path offset region is adjusted. If the update status of the path offset region of the adjusted cutting path curve meets the preset standard, a qualified cutting path curve is obtained. The screen cutting module is used to perform a cutting operation according to the qualified cutting path curve to obtain a finished flexible OLED screen. The final cutting path generation module is used to evaluate the edge quality of the finished flexible OLED screen and conduct electrical and optical performance tests on the finished screen. Through parameter adjustment and real-time feedback adjustment, a stable and matched final path is obtained.

Images