Three-dimensional Topography Laser Scanning Measurement and Reconstruction Method and System for Flexible Printed Circuit Board

Abstract

The present invention provides a method and system for laser scanning measurement and reconstruction of the three-dimensional topography of a flexible circuit board, which relates to the technical field of three-dimensional measurement. The method includes applying a preset stress field and constructing an array of strain monitoring points to obtain a strain distribution map of the flexible circuit board, and dividing the deformation-sensitive regions; arranging multiple groups of binocular laser scanning devices based on the distribution of the deformation-sensitive regions, collecting orthogonal grid projection images, and obtaining a spatial mapping relationship matrix through an adaptive multi-step calibration algorithm; combining the center line of the laser stripe to construct a depth information mapping model to generate initial three-dimensional point cloud data; performing point cloud registration and fusion through a local feature preservation algorithm and noise reduction and outlier removal through a local curvature adaptive algorithm to obtain optimized point cloud data; performing block fitting based on the deformation-sensitive regions, and optimizing and correcting the fitting result in combination with the strain distribution map and curvature distribution to generate the three-dimensional geometric topography of the flexible circuit board.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional measurement, and particularly relates to a method and system for laser scanning measurement and reconstruction of the three-dimensional topography of a flexible circuit board. Background Art

[0002] Due to its thin, light, and bendable characteristics, flexible circuit boards are widely used in fields such as smart wearables and aerospace. In practical applications, flexible circuit boards are prone to deformation, and the accurate measurement and reconstruction of their three-dimensional topography are of great significance for product quality control and reliability assessment. Currently, common three-dimensional topography measurement methods include contact measurement and non-contact measurement. Among them, non-contact measurement based on binocular laser scanning has gradually become the main method for measuring the three-dimensional topography of flexible circuit boards due to its non-destructive and efficient characteristics.

[0003] However, the existing measurement methods have the following deficiencies: Traditional scanning measurements adopt a fixed layout, which is difficult to adapt to the complex deformation characteristics of flexible circuit boards, resulting in measurement blind spots and data loss; Existing point cloud data processing methods do not consider the particularity of deformation-sensitive regions and adopt a unified noise reduction and reconstruction strategy, causing geometric feature distortion in key regions; Conventional surface fitting algorithms lack consideration of material mechanical properties and cannot ensure the consistency between the reconstruction result and the actual deformation state. These problems seriously affect the accuracy and reliability of the three-dimensional topography reconstruction of flexible circuit boards.

[0004] In summary, there is an urgent need for a method for laser scanning measurement and reconstruction of the three-dimensional topography of a flexible circuit board based on deformation-sensitive regions. By monitoring the strain distribution to determine the deformation-sensitive regions, optimizing the layout position of the laser scanning device, and adopting an adaptive exposure strategy to improve the quality of data acquisition; Combining local feature-preserving algorithms and curvature adaptive algorithms to optimize and process the point cloud data; Performing block fitting based on deformation-sensitive regions and optimizing and correcting it in combination with strain distribution information, high-precision reconstruction of the three-dimensional topography of flexible circuit boards is achieved. The present invention can solve the problems in the prior art. Summary of the Invention

[0005] Embodiments of the present invention provide a method and system for laser scanning measurement and reconstruction of the three-dimensional topography of a flexible circuit board, which can solve the problems in the prior art.

[0006] In the first aspect of the embodiments of the present invention,

[0007] A method for laser scanning measurement and reconstruction of the three-dimensional topography of a flexible circuit board is provided, including:

[0008] Apply a preset stress field to the flexible circuit board to be measured, construct a grid-like strain monitoring point array on the surface of the flexible circuit board, obtain the strain distribution map of the flexible circuit board through a multi-spectral camera, and divide the deformation-sensitive area based on the strain distribution map; calculate the optimal arrangement positions of four groups of binocular laser scanning devices based on the distribution of the deformation-sensitive area, and use an adaptive exposure control strategy to collect the orthogonal grid projection image group of the flexible circuit board;

[0009] Input the calibration data of the four groups of binocular laser scanning devices into an adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix; perform double Gaussian fitting on the orthogonal grid projection image group, extract the center line of the laser stripe, combine the center line of the laser stripe with the spatial mapping relationship matrix to construct a depth information mapping model, calculate the three-dimensional coordinates according to the depth information mapping model, generate initial three-dimensional point cloud data, and perform resampling and registration fusion on the initial three-dimensional point cloud data through a local feature preservation algorithm to obtain the fused point cloud data;

[0010] Input the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data; divide the surface of the flexible circuit board into multiple feature sub-regions according to the deformation-sensitive area, use an adaptive spline surface to perform block fitting on the feature sub-regions to determine the fitting result; calculate the curvature distribution of the flexible circuit board based on the optimized point cloud data, and optimize and correct the fitting result in combination with the strain distribution map to generate the three-dimensional geometric shape of the flexible circuit board.

[0011] In an optional embodiment,

[0012] Applying a preset stress field to the flexible circuit board to be measured, constructing a grid-like strain monitoring point array on the surface of the flexible circuit board, and dividing the deformation-sensitive area based on the strain distribution map obtained by a multi-spectral camera includes:

[0013] Apply a preset stress field to the flexible circuit board, construct a grid-like strain monitoring point array on the surface of the flexible circuit board, and obtain a stressed flexible circuit board with the grid-like strain monitoring point array; collect the multi-spectral images of the stressed flexible circuit board before and after applying the preset stress field, and perform registration processing on the multi-spectral images to obtain the registered deformation image;

[0014] The registered deformed image is divided into multiple sub-regions, and each of the sub-regions contains at least one monitoring point in the grid strain monitoring point array; a displacement correlation function is constructed between adjacent sub-regions, and the displacement correlation function includes the gray value distribution and displacement increment of the sub-region; the displacement correlation function is solved by an iterative optimization algorithm to obtain the displacement vector of the sub-region, and the displacement gradient of each sub-region is calculated based on the displacement vector to obtain the displacement field distribution of the flexible circuit board;

[0015] The displacement field distribution is interpolated using a preset radial basis function to obtain the continuous displacement field of the flexible circuit board. The strain tensor of the flexible circuit board is calculated based on the continuous displacement field to obtain the strain distribution map of the flexible circuit board. The principal strain amplitude is calculated according to the strain distribution map, and the region where the principal strain amplitude exceeds the preset strain threshold is determined as the candidate sensitive region. The strain gradient value within the candidate sensitive region is calculated to obtain the strain gradient value distribution map;

[0016] Based on the strain gradient value distribution map, the strain gradient maximum point is determined as the seed point within the candidate sensitive region, and a strain monitoring network is constructed with the seed point as the center. The strain monitoring network includes multiple concentric annular monitoring regions. The radius difference between adjacent concentric annular monitoring regions is determined according to the strain gradient value distribution map, and multiple strain monitoring nodes are arranged within each concentric annular monitoring region;

[0017] The correlation degree between the strain value of each strain monitoring node and the strain values of the strain monitoring nodes in the adjacent concentric annular monitoring region is calculated, and the strain weight value of each strain monitoring node is determined based on the correlation degree; the region corresponding to the strain monitoring node with the strain weight value greater than the preset weight threshold is determined as the deformation sensitive region of the flexible circuit board.

[0018] In an alternative embodiment,

[0019] Based on the distribution of the deformation sensitive region, calculating the optimal arrangement positions of four groups of binocular laser scanning devices includes:

[0020] The deformation sensitive region is represented as multiple planar sub-regions, and each of the planar sub-regions contains a central coordinate and an area weight. Based on the central coordinate and the area weight, a position parameter set of four groups of binocular laser scanning devices is constructed;

[0021] Calculate the observation angles of each group of the binocular laser scanning devices for each of the plane sub-regions, simultaneously evaluate the laser stripe imaging quality coefficients at each of the observation angles, and calculate the occlusion coefficients of each of the plane sub-regions at each of the observation angles; determine the visibility scores of the plane sub-regions based on the product of the observation angles, the imaging quality coefficients, and the occlusion coefficients, and construct an optimization objective function based on the area weights and the visibility scores;

[0022] Construct a first candidate solution group and a second candidate solution group based on the set of position parameters, where each candidate solution carries a set of the position parameter set; the first search radius corresponding to the first candidate solution group is greater than the second search radius corresponding to the second candidate solution group; calculate the optimization objective function values corresponding to each candidate solution; select the candidate solution with the highest optimization objective function value in the first candidate solution group as the first optimal solution, and select the candidate solution with the highest optimization objective function value in the second candidate solution group as the second optimal solution;

[0023] Centered on the first optimal solution, within the range of the first search radius, randomly generate a corresponding number of new first position parameter sets according to a preset first quantity, and update the candidate solutions carrying the new first position parameter sets to the first candidate solution group to obtain an updated first candidate solution group; centered on the second optimal solution, within the range of the second search radius, randomly generate a corresponding number of new second position parameter sets according to a preset second quantity, and update the candidate solutions carrying the new second position parameter sets to the second candidate solution group to obtain an updated second candidate solution group;

[0024] Calculate the optimization objective function values of each candidate solution in the updated first candidate solution group and the updated second candidate solution group; if the optimization objective function value of the first optimal solution is greater than the optimization objective function value of the second optimal solution, replace the candidate solution with the lowest optimization objective function value in the second candidate solution group with the position parameter set of the first optimal solution;

[0025] Repeat the iterative process, and based on a preset number of consecutive iterations, when the maximum improvement amplitude of the optimization objective function value is less than a preset improvement threshold, determine the position parameter set corresponding to the first optimal solution as the optimal layout scheme of the four groups of binocular laser scanning devices.

[0026] In an alternative embodiment,

[0027] Input the calibration data of the four groups of binocular laser scanning devices into an adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix, including:

[0028] Obtain the calibration image and calibration data of the binocular laser scanning device, extract the initial values of the internal parameters of the binocular laser scanning device based on the calibration data, where the initial values of the internal parameters include focal length parameters, principal point coordinate parameters, and radial distortion coefficient parameters; construct an internal camera parameter matrix based on the focal length parameters and the principal point coordinate parameters, and construct a distortion coefficient matrix based on the radial distortion coefficient parameters; combine the internal camera parameter matrix and the distortion coefficient matrix to construct an initial projection matrix;

[0029] Extract a set of projection points from the calibration image, divide the projection points in the set of projection points into a central region projection point subset and an edge region projection point subset according to the distance to the center point of the image; apply the initial projection matrix to the central region projection point subset and the edge region projection point subset respectively, and calculate the mean value and standard deviation of the reprojection error of each region;

[0030] Determine the region weight term according to the mean value and standard deviation of the reprojection error corresponding to the region where the projection point is located, determine the distance weight term according to the distance from the projection point to the center point of the image, and determine a dual adaptive weight factor based on the region weight term and the distance weight term;

[0031] Multiply the dual adaptive weight factor by the square of the reprojection error of the corresponding projection point to construct a weighted reprojection error, use the weighted reprojection error to construct an optimization objective function, and perform optimization calculations on the projection points of the central region projection point subset and the edge region projection point subset respectively based on the optimization objective function to update the initial values of the internal parameters and obtain an optimized internal parameter matrix;

[0032] Calculate the initial value of the rotation matrix based on the central region projection point subset, use the initial value of the rotation matrix as the initial value, and calculate the external parameter matrix in combination with the radial distortion characteristics of the edge region projection point subset, where the external parameter matrix includes the final rotation matrix parameter and the translation vector parameter; combine the optimized internal parameter matrix and the external parameter matrix to construct a spatial mapping relationship matrix;

[0033] Calculate the reprojection error of the projection points in the central region projection point subset and the edge region projection point subset respectively. When the reprojection error of the central region projection point subset is greater than the first preset threshold or the reprojection error of the edge region projection point subset is greater than the second preset threshold, update the dual adaptive weight factor according to the current reprojection error distribution and return for continued iteration until the reprojection error of the central region projection point subset is less than the first preset threshold and the reprojection error of the edge region projection point subset is less than the second preset threshold, and determine the current spatial mapping relationship matrix as the final spatial mapping relationship matrix.

[0034] In an alternative embodiment,

[0035] Perform double Gaussian fitting on the orthogonal grid projection image group, extract the center line of the laser stripe, combine the center line of the laser stripe with the spatial mapping relationship matrix to construct a depth information mapping model, calculate the three-dimensional coordinates according to the depth information mapping model, generate initial three-dimensional point cloud data, and perform resampling and registration fusion on the initial three-dimensional point cloud data through a local feature preservation algorithm. The fused point cloud data obtained includes:

[0036] Perform grayscale processing on the orthogonal grid projection image group to obtain grayscale images, decompose the grayscale images into multiple layers of sub-images through a multi-scale decomposition method, and determine corresponding adaptive windows based on the local grayscale gradient amplitudes of each layer of sub-images in the multiple layers of sub-images;

[0037] Collect sampling points of the laser stripe grayscale distribution within the adaptive window, perform outlier detection on the sampling points to obtain optimized sampling points, establish a double Gaussian distribution model based on the optimized sampling points, and the double Gaussian distribution model includes a first Gaussian distribution function and a second Gaussian distribution function; construct a weighted least squares objective function with a regularization term, and by minimizing the weighted least squares objective function, solve the parameters of the first Gaussian distribution function and the second Gaussian distribution function to obtain a multi-scale laser stripe grayscale distribution fitting model;

[0038] Extract the center line of the laser stripe of each layer of sub-images in the multiple layers of sub-images based on the multi-scale laser stripe grayscale distribution fitting model to obtain a multi-scale laser stripe center line, perform adaptive fusion on the multi-scale laser stripe center line to obtain a fused center line of the laser stripe, combine the fused center line of the laser stripe with the spatial mapping relationship matrix, establish a non-linear spatial mapping relationship, and calculate the spatial coordinate points corresponding to the fused center line of the laser stripe according to the non-linear spatial mapping relationship to obtain initial three-dimensional point cloud data;

[0039] Calculate the principal curvature parameter and the principal direction parameter of each spatial coordinate point in the initial three-dimensional point cloud data, calculate the point distribution entropy of the spatial coordinate points based on the principal curvature parameter and the principal direction parameter; combine the principal curvature parameter, the principal direction parameter and the point distribution entropy to construct a feature descriptor; calculate the adaptive weight coefficient of the spatial coordinate points based on the feature descriptor; perform hierarchical resampling on the initial three-dimensional point cloud data according to the adaptive weight coefficient to obtain resampled point cloud data;

[0040] Construct a local consistency constraint including point position constraint, normal vector constraint and curvature constraint, integrate the local consistency constraint into the iterative closest point algorithm, and perform registration and local fusion on the resampled point cloud data to obtain fused three-dimensional point cloud data.

[0041] Construct local consistency constraints including point constraints, normal vector constraints, and curvature constraints, integrate the local consistency constraints into the iterative closest point algorithm to obtain a constrained optimization model; perform iterative registration on the resampled point cloud data based on the constrained optimization model to obtain registered point cloud data; perform local fusion on the registered point cloud data according to the feature descriptor to obtain fused three-dimensional point cloud data.

[0042] In an alternative embodiment,

[0043] Inputting the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data includes:

[0044] Perform spatial segmentation on the fused point cloud data based on the Euclidean distance between points and the point cloud density to obtain multiple point cloud subsets, construct a KD-tree index structure for each point cloud subset, and calculate the nearest neighbor point set of each point based on the KD-tree index structure;

[0045] For each nearest neighbor point set, calculate the local structure tensor using a preset tensor voting algorithm, decompose the local structure tensor based on the eigenvalues to obtain the normal vector and tangent plane of each point; construct an orthogonal basis function family on the tangent plane, fit the local surface using the orthogonal basis function family, and calculate the first principal curvature and the second principal curvature of each point according to the local surface;

[0046] Calculate the variances of the first principal curvature and the second principal curvature of each point in the nearest neighbor point set, and when the variance is greater than the preset threshold, mark the corresponding point as a candidate outlier; calculate the spatial distribution density of the points in the nearest neighbor point set, and verify the candidate outliers based on the spatial distribution density to obtain an outlier set;

[0047] For the points that do not belong to the outlier set, construct a Gaussian curvature map based on the first principal curvature and the second principal curvature, calculate the local entropy value of the Gaussian curvature map, determine the local entropy value as the local feature intensity, calculate the shape parameter and scale parameter of the adaptive smoothing kernel function according to the local feature intensity, and construct an adaptive smoothing kernel function; apply the adaptive smoothing kernel function to the nearest neighbor point set, and calculate the contribution weight of each nearest neighbor point based on the Euclidean distance between points and the corresponding angle between the normal vectors, where the contribution weight decreases as the Euclidean distance between points increases and decreases as the corresponding angle between the normal vectors increases;

[0048] Construct a geometric feature preservation function, which includes a position constraint term and a normal vector consistency constraint term. The position constraint term calculates the weighted average deviation of point positions based on the contribution weights, and the normal vector consistency constraint term calculates the weighted average deviation of normal vectors based on the contribution weights. Update the positions and normal vectors of points by iteratively optimizing the geometric feature preservation function, and at the same time remove the points in the abnormal point set to obtain the optimized point cloud data.

[0049] In an optional embodiment,

[0050] Divide the surface of the flexible circuit board into multiple feature sub-regions according to the deformation-sensitive region, perform block fitting on the feature sub-regions using an adaptive spline surface, and determine the fitting result. Calculate the curvature distribution of the flexible circuit board based on the optimized point cloud data, and optimize and correct the fitting result in combination with the strain distribution map to generate the three-dimensional geometric morphology of the flexible circuit board, including:

[0051] Obtain the point cloud data on the surface of the flexible circuit board and the position data of the deformation-sensitive region on the surface of the flexible circuit board. Extract a set of feature-enhanced points based on the position data of the deformation-sensitive region, calculate the feature importance value of each point in the set of feature-enhanced points, and determine the mesh division parameters according to the feature importance value. The mesh division parameters decrease as the feature importance value increases.

[0052] Perform adaptive quadtree partitioning on the point cloud data using the mesh division parameters to obtain multiple feature sub-regions. Establish a local coordinate system in each feature sub-region, project the point cloud data in the feature sub-region onto the local coordinate system, and calculate the local curvature distribution of the projected point cloud data in the feature sub-region.

[0053] Construct a weight coefficient according to the local curvature distribution, and the weight coefficient increases as the local curvature distribution increases. Build an adaptive spline surface model based on the weight coefficient, and perform surface fitting on the feature sub-region through B-spline basis functions to obtain the initial fitting surface of the feature sub-region.

[0054] Obtain the strain distribution data of the flexible circuit board, calculate the Gaussian curvature value and the mean curvature value of the feature sub-region based on the initial fitting surface, and construct a surface correction function according to the strain distribution data, the Gaussian curvature value and the mean curvature value. The correction amount of the surface correction function is positively correlated with the strain distribution data.

[0055] Calculate the normal vector of the feature sub-region, apply the correction amount of the surface correction function along the direction of the normal vector to the initial fitting surface to obtain an optimized fitting surface; generate the three-dimensional geometric shape of the flexible circuit board by iteratively optimizing the parameters of the surface correction function to match the optimized fitting surface with the strain distribution data.

[0056] In the second aspect of the embodiments of the present invention,

[0057] A laser scanning measurement and reconstruction system for the three-dimensional topography of a flexible circuit board is provided, including:

[0058] A first unit for applying a preset stress field to the flexible circuit board to be measured, constructing a grid-like strain monitoring point array on the surface of the flexible circuit board, obtaining the strain distribution map of the flexible circuit board through a multi-spectral camera, and dividing the deformation-sensitive region based on the strain distribution map; calculating the optimal arrangement positions of four groups of binocular laser scanning devices based on the distribution of the deformation-sensitive region, and collecting an orthogonal grid projection image group of the flexible circuit board by using an adaptive exposure control strategy;

[0059] A second unit for inputting the calibration data of the four groups of binocular laser scanning devices into an adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix; performing double Gaussian fitting on the orthogonal grid projection image group, extracting the center line of the laser stripe, combining the center line of the laser stripe with the spatial mapping relationship matrix to construct a depth information mapping model, calculating three-dimensional coordinates according to the depth information mapping model, generating initial three-dimensional point cloud data, and performing resampling and registration fusion on the initial three-dimensional point cloud data through a local feature preservation algorithm to obtain fused point cloud data;

[0060] A third unit for inputting the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data; dividing the surface of the flexible circuit board into multiple feature sub-regions according to the deformation-sensitive region, performing block fitting on the feature sub-regions by using an adaptive spline surface to determine the fitting result; calculating the curvature distribution of the flexible circuit board based on the optimized point cloud data, and optimizing and correcting the fitting result in combination with the strain distribution map to generate the three-dimensional geometric shape of the flexible circuit board.

[0061] In the third aspect of the embodiments of the present invention,

[0062] An electronic device is provided, including:

[0063] A processor;

[0064] A memory for storing processor-executable instructions;

[0065] Wherein, the processor is configured to call the instructions stored in the memory to execute the method described above.

[0066] In a fourth aspect of the embodiments of the present invention,

[0067] a computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method described above is implemented.

[0068] In the embodiments of the present invention, by constructing a grid-shaped strain monitoring point array and a multi-spectral camera to obtain a strain distribution map, dividing the deformation-sensitive area, and optimizing the arrangement position of the binocular laser scanning device based on this, the subtle deformation of the flexible circuit board can be effectively captured, thereby improving the accuracy of three-dimensional topography measurement. The adaptive multi-step calibration algorithm and double Gaussian fitting further improve the accuracy of the data; by using four groups of binocular laser scanning devices and an adaptive exposure control strategy, an orthogonal grid projection image group can be quickly acquired, accelerating the speed of three-dimensional topography reconstruction; the application of the local feature preservation algorithm and the local curvature adaptive algorithm optimizes the point cloud data processing flow, further improving the measurement efficiency; by performing block fitting based on the deformation-sensitive area and combining the curvature distribution and the strain distribution map for optimization and correction, the complex three-dimensional topography of the flexible circuit board can be restored more accurately, especially in the area with large deformation; effectively avoiding problems such as error accumulation and distortion that occur in traditional methods when dealing with complex deformations, enhancing the integrity and robustness of topography reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 is a schematic flow chart of the method for laser scanning measurement and reconstruction of the three-dimensional topography of a flexible circuit board according to the embodiments of the present invention;

[0070] Figure 2 is a simulation diagram of observation angle quality evaluation and optimal layout;

[0071] Figure 3 is a comparison diagram of the simulation effects of three-dimensional point cloud reconstruction;

[0072] Figure 4 is a comparison broken line graph of the convergence performance of the surface fitting algorithm;

[0073] Figure 5 is a schematic structural diagram of the system for laser scanning measurement and reconstruction of the three-dimensional topography of a flexible circuit board according to the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0074] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part rather than all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0075] The technical solutions of the present invention will be described in detail below with specific embodiments. These specific embodiments may be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0076] Figure 1 It is a schematic flowchart of the method for three-dimensional topography laser scanning measurement and reconstruction of the flexible circuit board in the embodiment of the present invention. As Figure 1 shown, the method includes:

[0077] S101. Apply a preset stress field to the flexible circuit board to be measured, construct a grid-shaped strain monitoring point array on the surface of the flexible circuit board, obtain the strain distribution map of the flexible circuit board through a multispectral camera, and divide the deformation-sensitive area based on the strain distribution map; calculate the optimal arrangement positions of four groups of binocular laser scanning devices based on the distribution of the deformation-sensitive area, and collect an orthogonal grid projection image group of the flexible circuit board by using an adaptive exposure control strategy;

[0078] In this embodiment, by applying a preset stress field to the flexible circuit board and constructing a grid-shaped strain monitoring point array, the accurate measurement of its strain distribution is realized; the strain distribution map is obtained by using a multispectral camera and the deformation-sensitive area is divided to improve the recognition ability of key deformation areas; the arrangement positions of the binocular laser scanning devices are optimized based on the deformation-sensitive area to ensure the efficiency and accuracy of data collection; an orthogonal grid projection image group is obtained by using an adaptive exposure control strategy to improve the imaging quality and data stability, providing high-precision data support for subsequent three-dimensional reconstruction and topography analysis.

[0079] S102. Input the calibration data of the four groups of binocular laser scanning devices into an adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix; perform double Gaussian fitting on the orthogonal grid projection image group to extract the center line of the laser stripe, combine the center line of the laser stripe with the spatial mapping relationship matrix to construct a depth information mapping model, calculate three-dimensional coordinates according to the depth information mapping model, generate initial three-dimensional point cloud data, and perform resampling and registration fusion on the initial three-dimensional point cloud data through a local feature preservation algorithm to obtain fused point cloud data;

[0080] In this embodiment, the spatial mapping relationship of the binocular laser scanning device is accurately obtained through an adaptive multi-step calibration algorithm to improve the accuracy of three-dimensional measurement; double Gaussian fitting is used to extract the center line of the laser stripe to enhance the accuracy and robustness of feature point extraction; a depth information mapping model is used to calculate three-dimensional coordinates and generate initial point cloud data to achieve efficient three-dimensional reconstruction; a local feature preservation algorithm is used for resampling and registration fusion to effectively improve the integrity and consistency of the point cloud data, providing a high-quality data basis for subsequent geometric analysis and topography optimization.

[0081] S103. Input the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data; divide the surface of the flexible circuit board into multiple feature sub-regions according to the deformation-sensitive region, and use an adaptive spline surface to perform piecewise fitting on the feature sub-regions to determine the fitting result; calculate the curvature distribution of the flexible circuit board based on the optimized point cloud data, and optimize and correct the fitting result in combination with the strain distribution map to generate the three-dimensional geometry of the flexible circuit board.

[0082] In this embodiment, noise reduction and outlier removal are performed on the fused point cloud data through a local curvature adaptive algorithm to improve the purity and reliability of the point cloud data; feature division is performed on the surface of the flexible circuit board based on the deformation-sensitive region, and an adaptive spline surface is used for piecewise fitting to ensure accurate description of complex topography; the curvature distribution is calculated using the optimized point cloud data, and the fitting result is optimized and corrected in combination with the strain distribution map, thereby improving the accuracy of three-dimensional geometry reconstruction and providing high-precision data support for the quality inspection and performance evaluation of the flexible circuit board.

[0083] In an alternative embodiment, a preset stress field is applied to the flexible circuit board to be measured, a grid-like strain monitoring point array is constructed on the surface of the flexible circuit board, and a strain distribution map of the flexible circuit board is obtained through a multi-spectral camera. Dividing the deformation-sensitive region based on the strain distribution map includes:

[0084] Apply a preset stress field to the flexible circuit board, construct a grid-like strain monitoring point array on the surface of the flexible circuit board, and obtain a stressed flexible circuit board with the grid-like strain monitoring point array; collect multi-spectral images of the stressed flexible circuit board before and after applying the preset stress field, and perform registration processing on the multi-spectral images to obtain a registered deformation image;

[0085] The registered deformed image is divided into multiple sub-regions, and each of the sub-regions contains at least one monitoring point in the grid-like strain monitoring point array; a displacement correlation function is constructed between adjacent sub-regions, and the displacement correlation function includes the gray value distribution and displacement increment of the sub-region; the displacement correlation function is solved by an iterative optimization algorithm to obtain the displacement vector of the sub-region, and the displacement gradient of each sub-region is calculated based on the displacement vector to obtain the displacement field distribution of the flexible circuit board;

[0086] The displacement field distribution is interpolated by a preset radial basis function to obtain the continuous displacement field of the flexible circuit board, the strain tensor of the flexible circuit board is calculated based on the continuous displacement field to obtain the strain distribution map of the flexible circuit board, the principal strain amplitude is calculated according to the strain distribution map, the region where the principal strain amplitude exceeds the preset strain threshold is determined as the candidate sensitive region, the strain gradient value in the candidate sensitive region is calculated to obtain the strain gradient value distribution map;

[0087] Based on the strain gradient value distribution map, the strain gradient maximum point is determined as the seed point in the candidate sensitive region, and a strain monitoring network is constructed with the seed point as the center, where the strain monitoring network includes multiple concentric annular monitoring regions, the radius difference between adjacent concentric annular monitoring regions is determined according to the strain gradient value distribution map, and multiple strain monitoring nodes are set in each concentric annular monitoring region;

[0088] The correlation degree between the strain value of each strain monitoring node and the strain values of the strain monitoring nodes in its adjacent concentric annular monitoring region is calculated, and the strain weight value of each strain monitoring node is determined based on the correlation degree; the region corresponding to the strain monitoring node with the strain weight value greater than the preset weight threshold is determined as the deformation sensitive region of the flexible circuit board.

[0089] In a specific embodiment, first, a preset stress field needs to be applied to the flexible circuit board and a strain monitoring point array is constructed, the deformation information is obtained through multi-spectral image acquisition, and then the accurate identification of the deformation sensitive region is realized through displacement field analysis and strain monitoring network construction.

[0090] When constructing the strain monitoring point array on the surface of the flexible circuit board, a photosensitive material is coated on the surface by a lithography process, and regular grid-like monitoring points are made through exposure and development, with a grid spacing of 2 mm, covering the entire surface of the flexible circuit board. A customized multi-degree-of-freedom load application device is used to apply a preset stress field to the flexible circuit board, a tensile force of 5 N is applied at one end, and a torque of 0.5 N·m is applied in the middle at the same time to cause the flexible circuit board to produce composite deformation.

[0091] Use a multispectral camera equipped with three bands of 540 nm, 680 nm, and 820 nm to collect image groups before and after deformation respectively. Preprocess the collected images, including lens distortion correction and dark field correction, to eliminate systematic errors and random noise. Register the images before and after deformation through feature point matching method to obtain a registered image group that accurately reflects the deformation information.

[0092] Divide the registered images into overlapping sub-region grids with a sub-region size of 5×5 mm and an overlap rate of 50% between adjacent sub-regions. Establish a correlation analysis model considering the gray value distribution and displacement increment between adjacent sub-regions. Use the iterative optimization method to solve the correlation function to obtain the displacement vector of the sub-region. Calculate the displacement gradient field based on the displacement vector to obtain the displacement field distribution reflecting the overall deformation characteristics.

[0093] Perform radial basis function interpolation on the displacement field distribution to obtain a continuous displacement field function. Calculate each component of the strain tensor based on the continuous displacement field to obtain the strain distribution map. Extract the principal strain direction and amplitude, and mark the area where the principal strain amplitude exceeds 3% as the candidate sensitive area. Calculate the strain gradient value within the candidate sensitive area to obtain the strain gradient distribution map.

[0094] Find the local maximum points in the strain gradient distribution map as seed points, and construct three-layer concentric annular monitoring areas centered on the seed points with radii of 3 mm, 6 mm, and 9 mm respectively. Determine the spacing between adjacent annular areas according to the change law of the strain gradient, and evenly arrange 12 monitoring nodes on each annular area.

[0095] Calculate the strain correlation between monitoring nodes, and use the weight evaluation method based on the strain gradient to determine the strain weight value of each monitoring node. Determine the area where the monitoring node with a strain weight value greater than 0.8 is located as the deformation sensitive area.

[0096] In practical applications, taking a flexible circuit board for a smart bracelet as an example, three main deformation sensitive areas located in the bending area and near the solder joints of the device are successfully identified through the above method. The maximum strain gradient value of these areas reaches 0.15 / mm, and the strain weight value exceeds 0.85, indicating that these areas are the positions with the most significant deformation. The identification result of the deformation sensitive area provides an important basis for subsequent three-dimensional shape reconstruction and effectively improves the reconstruction accuracy.

[0097] This method based on multispectral image analysis and strain monitoring network realizes the accurate identification of deformation sensitive areas through multi-level strain information extraction and correlation analysis, and provides effective technical support for the shape reconstruction and reliability evaluation of flexible circuit boards.

[0098] In this embodiment, by constructing a strain monitoring network and calculating strain weight values, the deformation-sensitive areas of the flexible circuit board can be identified more accurately, avoiding the errors caused by relying solely on strain amplitudes for judgment; through multispectral imaging and displacement field calculation, a high-resolution strain distribution map can be obtained, thereby realizing the refined division of the deformation-sensitive areas and better guiding the design and manufacture of the flexible circuit board; by identifying and monitoring the deformation-sensitive areas, the failure positions of the flexible circuit board can be predicted in advance and corresponding measures can be taken, thereby improving the reliability and service life of the flexible circuit board.

[0099] In an alternative embodiment, calculating the optimal arrangement positions of the four groups of binocular laser scanning devices based on the distribution of the deformation-sensitive areas includes:

[0100] Represent the deformation-sensitive areas as multiple planar sub-regions, each of the planar sub-regions including a central coordinate and an area weight, and construct a set of position parameters for the four groups of binocular laser scanning devices based on the central coordinate and the area weight;

[0101] Calculate the observation angles of each group of binocular laser scanning devices for each of the planar sub-regions, and simultaneously evaluate the laser stripe imaging quality coefficient at each of the observation angles, and calculate the occlusion coefficient of each of the planar sub-regions at each of the observation angles; determine the visibility score of the planar sub-region based on the product of the observation angle, the imaging quality coefficient, and the occlusion coefficient, and construct an optimization objective function based on the area weight and the visibility score;

[0102] Construct a first candidate solution group and a second candidate solution group based on the set of position parameters, each of the candidate solutions carrying a set of the position parameter sets; the first search radius corresponding to the first candidate solution group is greater than the second search radius corresponding to the second candidate solution group; calculate the optimization objective function value corresponding to each candidate solution; select the candidate solution with the highest optimization objective function value in the first candidate solution group as the first optimal solution, and select the candidate solution with the highest optimization objective function value in the second candidate solution group as the second optimal solution;

[0103] Centered on the first optimal solution, within the range of the first search radius, randomly generate a corresponding number of new first position parameter sets according to a preset first quantity, and update the candidate solutions carrying the new first position parameter sets to the first candidate solution group to obtain an updated first candidate solution group; centered on the second optimal solution, within the range of the second search radius, randomly generate a corresponding number of new second position parameter sets according to a preset second quantity, and update the candidate solutions carrying the new second position parameter sets to the second candidate solution group to obtain an updated second candidate solution group;

[0104] Calculate the optimized objective function values of each candidate solution in the updated first candidate solution group and the updated second candidate solution group; if the optimized objective function value of the first optimal solution is greater than that of the second optimal solution, replace the candidate solution with the lowest optimized objective function value in the second candidate solution group with the position parameter set of the first optimal solution;

[0105] Repeat the iterative process. In the continuous iteration based on a preset number of times, when the maximum improvement amplitude of the optimized objective function value is less than the preset improvement threshold, determine the position parameter set corresponding to the first optimal solution as the optimal layout scheme of the four groups of binocular laser scanning devices.

[0106] In a specific embodiment, first, divide the deformation-sensitive area to be measured into multiple plane sub-regions. Each sub-region is represented by its central coordinates and area weight. The area weight reflects the importance of this sub-region in the overall deformation monitoring. For example, sub-regions with larger areas or sub-regions at key positions will be given greater weights. For example, a 10-meter × 10-meter deformation-sensitive area can be divided into 100 1-meter × 1-meter sub-regions. The central coordinates of each sub-region are its geometric center, and the area weights can be set to the same value or different values according to the actual situation. For example, sub-regions in the central area can be set with larger weights.

[0107] Next, construct the position parameter sets of the four groups of binocular laser scanning devices. Each group of binocular laser scanning devices includes two laser emitters and two receivers, and its position parameters include three-dimensional coordinates and rotation angles, etc. These parameters will be used as variables in the subsequent optimization process. The initial position parameter sets can be determined by experience or preliminary rough estimation. For example, the four groups of devices can be evenly distributed around the deformation-sensitive area and roughly oriented towards the center of the area.

[0108] Then, for each sub-region, calculate the observation angle of each group of binocular laser scanning devices with respect to it. The observation angle refers to the angle between the laser beam and the normal of the sub-region. At the same time, evaluate the laser stripe imaging quality coefficient at each observation angle. This coefficient reflects the clarity and stability of the laser stripe at this angle, and its value range is from 0 to 1. The larger the value, the better the imaging quality. The imaging quality coefficient can be obtained through simulation or experiments. For example, a laser with a perpendicular incidence usually has better imaging quality, and its imaging quality coefficient can be set to 1; while a laser with a larger tilt angle may be affected by scattering and reflection, and its imaging quality coefficient will be correspondingly reduced, such as set to 0.5 or lower. In addition, it is also necessary to calculate the occlusion coefficient of each sub-region at each observation angle. The occlusion coefficient represents the degree to which this sub-region is occluded by other objects at this angle, and its value range is also from 0 to 1. The larger the value, the more serious the occlusion. The occlusion coefficient can be calculated by establishing a three-dimensional model and performing ray tracing.

[0109] Based on the product of the observation angle, the imaging quality coefficient, and the occlusion coefficient, determine the visibility score of each sub-region. The higher the visibility score, the better the deformation monitoring effect of the sub-region at that observation angle. Then, based on the area weight and visibility score of each sub-region, construct an optimization objective function. The objective of the optimization objective function is to maximize the sum of the weighted visibility scores of all sub-regions.

[0110] To find the optimal device layout scheme, adopt a two-layer iterative optimization strategy. First, construct a first candidate solution group and a second candidate solution group. Each candidate solution carries a set of position parameter sets. The first search radius corresponding to the first candidate solution group is greater than the second search radius corresponding to the second candidate solution group. Calculate the optimization objective function value corresponding to each candidate solution. Select the candidate solution with the highest optimization objective function value in the first candidate solution group as the first optimal solution, and select the candidate solution with the highest optimization objective function value in the second candidate solution group as the second optimal solution.

[0111] Then, enter the iterative process. Centered on the first optimal solution, within the range of the first search radius, randomly generate a preset number of new position parameter sets, and update these new candidate solutions to the first candidate solution group. Similarly, centered on the second optimal solution, within the range of the second search radius, randomly generate a preset number of new position parameter sets, and update these new candidate solutions to the second candidate solution group. Calculate the optimization objective function values of each candidate solution in the two updated candidate solution groups. If the optimization objective function value of the first optimal solution is greater than that of the second optimal solution, replace the candidate solution with the lowest optimization objective function value in the second candidate solution group with the position parameter set of the first optimal solution. Repeat the above iterative process. When the maximum improvement amplitude of the optimization objective function value is less than the preset improvement threshold in a preset number of consecutive iterations, determine the position parameter set corresponding to the first optimal solution as the optimal layout scheme of the four binocular laser scanning devices. For example, set the number of iterations to 100 times and the improvement threshold to 0.01. If the improvement amplitude of the optimization objective function value does not exceed 0.01 in 100 consecutive iterations, it is considered that the algorithm has converged, and the first optimal solution at this time is the final optimization result.

[0112] The Figure 2Shows the optimal layout results of the binocular laser scanning device obtained based on the proposed technical solution. The figure clearly shows that the deformation-sensitive area is divided into 45 planar sub-areas (9×5 grid), and each sub-area contains central coordinate and area weight information (such as (5,4) and W = 0.72, etc.). Four groups of binocular laser scanning devices (Device A, B, C, D) are respectively arranged at the four corner positions of the observation space, with coordinates (100,100), (650,100), (650,450) and (100,450). Each device is marked with 4 main observation angles, represented by red rays. The observation angles of Device A are 28°, 45°, 65° and 83°, Device B are 115°, 135°, 152° and 175°, Device C are 195°, 215°, 232° and 255°, and Device D are 285°, 302°, 322° and 345°. These angles are calculated by an optimization algorithm to ensure the best coverage of the deformation-sensitive area. The simulation results show that this layout scheme achieves a comprehensive visibility score of 94.7%, the average occlusion coefficient is only 0.15, and the observation quality coefficients of each device are 0.92, 0.89, 0.95 and 0.88 respectively, indicating that this scheme can effectively overcome the occlusion problem and provide high-quality laser stripe imaging.

[0113] In this embodiment, by optimizing the arrangement positions of the binocular laser scanning devices, the visibility score of the deformation-sensitive area can be maximized, thereby improving the accuracy and reliability of deformation measurement; the optimal device arrangement scheme can be automatically determined, avoiding the cumbersome process of manual trial and error, thereby improving work efficiency; the device arrangement scheme can be flexibly adjusted according to the shape and distribution characteristics of different deformation-sensitive areas, thereby enhancing the adaptability of the method.

[0114] In an alternative embodiment, the calibration data of the four groups of binocular laser scanning devices are input into an adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix, including:

[0115] Obtain the calibration images and calibration data of the binocular laser scanning device, extract the initial values of the internal parameters of the binocular laser scanning device based on the calibration data, and the initial values of the internal parameters include focal length parameter, principal point coordinate parameter and radial distortion coefficient parameter; construct a camera internal parameter matrix based on the focal length parameter and the principal point coordinate parameter, and construct a distortion coefficient matrix based on the radial distortion coefficient parameter; combine the camera internal parameter matrix and the distortion coefficient matrix to construct an initial projection matrix;

[0116] Extract a set of projection points from the calibration image, and divide the projection points in the set of projection points into a central region projection point subset and a peripheral region projection point subset according to the distance to the center point of the image; respectively apply the initial projection matrix to the central region projection point subset and the peripheral region projection point subset, and calculate the mean value and standard deviation of the reprojection error of each region;

[0117] Determine the region weight term according to the mean value and standard deviation of the reprojection error corresponding to the region where the projection point is located, determine the distance weight term according to the distance from the projection point to the center point of the image, and determine the dual adaptive weight factor based on the region weight term and the distance weight term;

[0118] Multiply the dual adaptive weight factor by the square of the reprojection error of the corresponding projection point to construct a weighted reprojection error, use the weighted reprojection error to construct an optimization objective function, and perform optimization calculations on the projection points of the central region projection point subset and the peripheral region projection point subset respectively based on the optimization objective function, and update the initial value of the internal parameters to obtain an optimized internal parameter matrix;

[0119] Calculate the initial value of the rotation matrix based on the central region projection point subset, use the initial value of the rotation matrix as the initial value, and calculate the external parameter matrix in combination with the radial distortion characteristics of the peripheral region projection point subset. The external parameter matrix includes the final rotation matrix parameters and translation vector parameters; combine the optimized internal parameter matrix and the external parameter matrix to construct a spatial mapping relationship matrix;

[0120] Calculate the reprojection errors of the projection points in the central region projection point subset and the peripheral region projection point subset respectively. When the reprojection error of the central region projection point subset is greater than the first preset threshold or the reprojection error of the peripheral region projection point subset is greater than the second preset threshold, update the dual adaptive weight factor according to the current reprojection error distribution and return for continued iteration until the reprojection error of the central region projection point subset is less than the first preset threshold and the reprojection error of the peripheral region projection point subset is less than the second preset threshold, and determine the current spatial mapping relationship matrix as the final spatial mapping relationship matrix.

[0121] The calibration image specifically refers to a specific image used for camera calibration. Usually, a calibration board with a known structure or a laser scanning projection pattern is photographed to estimate the internal and external parameters of the camera and achieve accurate geometric measurement and three-dimensional reconstruction. The calibration image usually contains regular patterns, such as checkerboards, dot arrays or specific markers, in order to extract feature points for calculation. The main function of the calibration image is to obtain pixel coordinate information and cover different regions to evaluate the distortion and projection error within the entire field of view.

[0122] The calibration data specifically refers to the geometric and optical information extracted from the calibration images, which is used to calculate the parameters of the camera or laser scanning device to ensure the accuracy of measurement. The calibration data mainly includes internal parameters, external parameters, projection point data, and error analysis data. The internal parameters describe the internal optical characteristics of the camera, such as focal length, principal point coordinates, and distortion coefficients. The external parameters are used to represent the position and orientation of the camera in space, including the rotation matrix and translation vector. The projection point data is the coordinate of the feature points extracted from the calibration images, which is used to calculate the projection error, while the error analysis data is used to optimize the camera parameters and improve the measurement accuracy.

[0123] In a specific implementation, the calibration images and calibration data of the binocular laser scanning device are obtained. Taking the checkerboard calibration board as an example, multiple groups of calibration board images with different postures and positions are taken, and using the known size information of the calibration board, the pixel coordinates of the feature points in the images are extracted as the calibration data. For example, 15 groups of calibration board images are taken, with the resolution of each group of images being 1920×1080 pixels and the grid point size of the calibration board being 25mm.

[0124] Based on the calibration data, the initial values of the internal parameters of the binocular laser scanning device are extracted. Using the Zhang Zhengyou calibration method, based on the extracted calibration data, the initial values of the internal parameters of the binocular laser scanning device are calculated, including the focal length parameter, principal point coordinate parameter, and radial distortion coefficient parameter. For example, the initial value of the focal length of the left camera is (fx = 1200, fy = 1200), the initial value of the principal point coordinates is (cx = 960, cy = 540), and the initial value of the radial distortion coefficient is (k1 = 0.001, k2 = -0.0005).

[0125] The camera internal parameter matrix and the distortion coefficient matrix are constructed, and the two are combined to construct the initial projection matrix. According to the obtained focal length parameter and principal point coordinate parameter, the camera internal parameter matrix is constructed, and according to the radial distortion coefficient parameter, the distortion coefficient matrix is constructed. The camera internal parameter matrix and the distortion coefficient matrix are multiplied to obtain the initial projection matrix.

[0126] The projection point set is extracted from the calibration images and divided into a central region projection point subset and an edge region projection point subset. The pixel coordinates of all feature points in the calibration images are extracted to form the projection point set. The distance from each projection point to the center point of the image is calculated, and the projection points with a distance less than half of the distance from the center point of the image to the edge of the image are divided into the central region projection point subset, and the remaining projection points are divided into the edge region projection point subset. Assuming the center point coordinates of the image are (960, 540) and the image size is 1920×1080, then the radius of the central region is sqrt((960 / 2) 2 +(540 / 2) 2 )≈540.

[0127] Calculate the mean and standard deviation of the reprojection error for each region. Apply the initial projection matrix to the subset of projection points in the central region and the subset of projection points in the edge region respectively, and calculate the reprojection error for each projection point, that is, the difference between the actual pixel coordinates of the projection point and the pixel coordinates of the projection point calculated according to the projection matrix. Calculate the mean and standard deviation of the reprojection error for the central region and the edge region respectively. For example, the mean reprojection error for the central region is 0.5 pixels and the standard deviation is 0.2 pixels; the mean reprojection error for the edge region is 1.2 pixels and the standard deviation is 0.5 pixels.

[0128] Determine the dual adaptive weight factor according to the reprojection error and the distance from the projection point to the center point of the image. The regional weight term is calculated based on the mean and standard deviation of the regional reprojection error. For example, the regional weight term = 1 / (mean + standard deviation). The distance weight term is calculated according to the distance from the projection point to the center point of the image. For example, the distance weight term = 1 / (1 + distance / image diagonal length). The dual adaptive weight factor is the product of the regional weight term and the distance weight term.

[0129] Construct the weighted reprojection error sum and the optimization objective function, and perform optimization calculations to update the intrinsic parameter matrix. Multiply the dual adaptive weight factor by the square of the reprojection error of the corresponding projection point to obtain the weighted reprojection error. Take the sum of the weighted reprojection errors of all projection points as the optimization objective function. Use a non-linear optimization algorithm (such as the Levenberg-Marquardt algorithm) to minimize the optimization objective function, update the initial value of the intrinsic parameters, and obtain the optimized intrinsic parameter matrix.

[0130] Calculate the initial value of the rotation matrix based on the subset of projection points in the central region, and calculate the extrinsic parameter matrix in combination with the subset of projection points in the edge region. Use the optimized intrinsic parameter matrix and the subset of projection points in the central region to calculate the initial value of the rotation matrix by solving the pose estimation problem, such as using the PnP algorithm. Take the initial value of the rotation matrix as the initial value, and combine the radial distortion characteristics of the subset of projection points in the edge region, and use a non-linear optimization algorithm, such as the Bundle Adjustment algorithm, to calculate the extrinsic parameter matrix, including the final rotation matrix parameters and translation vector parameters.

[0131] Iteratively optimize the spatial mapping relationship matrix until the preset conditions are met. Calculate the reprojection errors of the projection points in the central region projection point subset and the edge region projection point subset respectively. Set a first preset threshold and a second preset threshold, for example, 0.1 pixel and 0.2 pixel respectively. When the reprojection error of the central region projection point subset is greater than the first preset threshold or the reprojection error of the edge region projection point subset is greater than the second preset threshold, update the dual adaptive weight factor according to the current reprojection error distribution, and return to recalculate the weighted projection error, optimize the internal parameters and external parameters until the reprojection errors of both regions are less than the corresponding preset thresholds. At this time, determine the current spatial mapping relationship matrix as the final spatial mapping relationship matrix.

[0132] In this embodiment, through the dual adaptive weight factor, the influence of the projection points in the edge region and the projection points with large distortion on the calibration result can be effectively reduced, thereby improving the calibration accuracy; the requirements for the quality of the calibration image and the pose of the calibration board are not high, and it has strong robustness; through regional optimization and the adaptive weight factor, the convergence speed of the calibration algorithm can be accelerated and the calibration time can be shortened.

[0133] In an alternative embodiment, perform dual Gaussian fitting on the orthogonal grid projection image group, extract the center line of the laser stripe, combine the center line of the laser stripe with the spatial mapping relationship matrix to construct a depth information mapping model, calculate the three-dimensional coordinates according to the depth information mapping model, generate the initial three-dimensional point cloud data, and perform resampling and registration fusion on the initial three-dimensional point cloud data through a local feature preservation algorithm. The fused point cloud data obtained includes:

[0134] Perform grayscale processing on the orthogonal grid projection image group to obtain a grayscale image, decompose the grayscale image into multiple layers of sub-images through a multi-scale decomposition method, and determine the corresponding adaptive window based on the local grayscale gradient amplitude of each layer of sub-images in the multiple layers of sub-images;

[0135] Collect sampling points of the laser stripe grayscale distribution within the adaptive window, perform outlier detection on the sampling points to obtain optimized sampling points, establish a dual Gaussian distribution model based on the optimized sampling points, and the dual Gaussian distribution model includes a first Gaussian distribution function and a second Gaussian distribution function; construct a weighted least squares objective function with a regularization term, and solve the parameters of the first Gaussian distribution function and the second Gaussian distribution function by minimizing the weighted least squares objective function to obtain a multi-scale laser stripe grayscale distribution fitting model; <{

[0136] Extract the laser stripe centerlines of each sub-image in the multi-layer sub-images based on the multi-scale laser stripe gray distribution fitting model to obtain multi-scale laser stripe centerlines, adaptively fuse the multi-scale laser stripe centerlines to obtain a fused laser stripe centerline, combine the fused laser stripe centerline with the spatial mapping relationship matrix to establish a non-linear spatial mapping relationship, and calculate the spatial coordinate points corresponding to the fused laser stripe centerline according to the non-linear spatial mapping relationship to obtain initial three-dimensional point cloud data;

[0137] Calculate the principal curvature parameter and principal direction parameter of each spatial coordinate point in the initial three-dimensional point cloud data, and calculate the point distribution entropy of the spatial coordinate point based on the principal curvature parameter and the principal direction parameter; combine the principal curvature parameter, the principal direction parameter, and the point distribution entropy to construct a feature descriptor; calculate the adaptive weight coefficient of the spatial coordinate point based on the feature descriptor; perform hierarchical resampling on the initial three-dimensional point cloud data according to the adaptive weight coefficient to obtain resampled point cloud data;

[0138] Construct a local consistency constraint including point position constraint, normal vector constraint, and curvature constraint, integrate the local consistency constraint into the iterative closest point algorithm to obtain a constraint optimization model; perform iterative registration on the resampled point cloud data based on the constraint optimization model to obtain registered point cloud data; perform local fusion on the registered point cloud data according to the feature descriptor to obtain fused three-dimensional point cloud data.

[0139] In a specific embodiment, first obtain a group of orthogonal grid projection images. This image group contains orthogonal grid patterns projected onto the surface of the object to be measured at different phases. Take a group of orthogonal grid projection images containing 10 pictures as an example.

[0140] Then perform grayscale processing on the obtained image group. Convert the color image to a grayscale image for subsequent processing. For example, use the weighted average method to convert RGB values to grayscale values.

[0141] Next, use a multi-scale decomposition method to process the grayscale image. For example, use wavelet transform to decompose the grayscale image into multi-layer sub-images, assuming it is decomposed into 3 layers. This can process image features at different scales respectively and improve the robustness of the algorithm.

[0142] For each layer of sub-image, adaptively determine the window size according to the local gray gradient amplitude. In areas with a larger gradient amplitude, the window size is smaller, and vice versa. For example, in areas where the gradient amplitude is between 0 - 10, the window size is 5×5; in areas where the gradient amplitude is between 10 - 50, the window size is 3×3; in areas where the gradient amplitude is greater than 50, the window size is 1×1.

[0143] Collect sampling points of the gray-scale distribution of the laser stripe within the adaptive window. For example, within a 5×5 window, collect the gray-scale values of 25 sampling points.

[0144] Perform outlier detection on the collected sampling points to remove noise points. For example, use a statistics-based method to consider points that deviate from the mean by more than three standard deviations as outliers and remove them.

[0145] Use the remaining optimized sampling points to establish a double Gaussian distribution model. This model consists of two Gaussian functions that respectively fit the bright and dark stripes of the laser stripe.

[0146] By minimizing the weighted least squares objective function, solve the parameters of the two Gaussian functions to obtain a multi-scale laser stripe gray-scale distribution fitting model. The objective function contains a regularization term to prevent overfitting.

[0147] Extract the centerlines of the laser stripes of each sub-image based on the fitting model to obtain multi-scale laser stripe centerlines. The position of the centerline corresponds to the center point between the peaks of the two Gaussian functions.

[0148] Perform adaptive fusion on the multi-scale laser stripe centerlines to obtain the fused laser stripe centerline. The fusion process takes into account the weights of the centerlines at different scales. For example, use a confidence-based weighted average method.

[0149] Combine the fused laser stripe centerline with the spatial mapping relationship matrix to establish a non-linear spatial mapping relationship. The spatial mapping relationship matrix describes the transformation relationship between the camera coordinate system and the projector coordinate system.

[0150] Calculate the spatial coordinate points corresponding to the fused laser stripe centerline according to the non-linear spatial mapping relationship to obtain the initial 3D point cloud data.

[0151] Calculate the principal curvature parameter and the principal direction parameter of each spatial coordinate point in the initial 3D point cloud data. For example, use the method of local surface fitting to calculate these parameters.

[0152] Calculate the point distribution entropy of the spatial coordinate points based on the principal curvature parameter and the principal direction parameter. The point distribution entropy reflects the distribution uniformity of the local point cloud.

[0153] Combine the principal curvature parameter, the principal direction parameter, and the point distribution entropy to construct a feature descriptor. This descriptor is used to characterize the local geometric features of each point.

[0154] Calculate the adaptive weight coefficient of the spatial coordinate points based on the feature descriptor. For example, points with larger feature differences are assigned larger weights.

[0155] The initial three-dimensional point cloud data is resampled layer by layer according to the adaptive weight coefficient to obtain the resampled point cloud data. The important feature points are retained during the resampling process.

[0156] Construct local consistency constraints including point position constraints, normal vector constraints, and curvature constraints. These constraints are used to ensure that the registered point cloud data maintains the consistency of local geometric features.

[0157] Integrate the local consistency constraints into the iterative closest point algorithm to obtain a constrained optimization model.

[0158] Based on the constrained optimization model, perform iterative registration on the resampled point cloud data to obtain the registered point cloud data.

[0159] Perform local fusion on the registered point cloud data according to the feature descriptor to obtain the final fused three-dimensional point cloud data. The fusion process considers feature similarity and merges similar points.

[0160] Figure 3 The comparison chart of the three-dimensional point cloud reconstruction simulation effects fully demonstrates the effect differences between the present technical solution and three existing technologies (binocular stereo vision method, traditional structured light method, and single Gaussian fitting method) when dealing with the same complex surface. In the figure, the results of the four methods are displayed side by side, and the complete key performance indicators are marked for each method. From the visual effect, the point cloud generated by the present technical solution is significantly denser and more evenly distributed, with a point cloud density of 1254 points / cm², much higher than other methods. Especially at the edges and details of the surface, the point distribution of the present technical solution is more accurate, which is consistent with its surface detail retention rate of 92.7% and edge sharpness retention rate of 89.6%. In contrast, the point cloud of the binocular stereo vision method is significantly sparser, with a point cloud density of only 743 points / cm², and the point distribution in the edge area is chaotic. Its surface detail retention rate is only 68.3%, and the edge sharpness retention rate is even lower to 57.2%, resulting in a reconstruction accuracy error as high as 0.246 mm.

[0161] Although the traditional structured light method performs better than the binocular method, with a point cloud density of 895 points / cm², a surface detail retention rate of 76.5%, an edge sharpness retention rate of 69.1%, and a comprehensive score of 7.67, its reconstruction accuracy error of 0.157 mm is still significantly higher than 0.075 mm of this technical solution. The single Gaussian fitting method is between the traditional structured light method and the binocular stereo vision method, with a point cloud density of 862 points / cm², a surface detail retention rate of 72.4%, an edge sharpness retention rate of 64.8%, a reconstruction accuracy error of 0.138 mm, and a comprehensive score of 7.29. It can also be clearly seen from the figure that this technical solution not only has an advantage in point cloud density, but also demonstrates its excellent ability to retain the geometric features of objects through the complete presentation of feature lines, which is consistent with its high point cloud integrity of 95.3% and comprehensive quality score of 9.34. Through the intuitive comparison of each method on the same complex surface and the complete presentation of all key indicators, this figure strongly proves the comprehensive performance advantages of this technical solution achieved through double Gaussian fitting and local feature preservation algorithm.

[0162] In this embodiment, through double Gaussian fitting and multi-scale processing, the center line of the laser stripe can be extracted more accurately, thereby improving the accuracy of 3D reconstruction; the point cloud resampling and registration fusion strategy based on local features can effectively remove noise points and redundant points and retain important feature points, thereby improving the integrity of point cloud data; the application of strategies such as adaptive window, adaptive fusion, and local consistency constraint makes this method more robust to noise and complex scenes and can obtain more stable and reliable 3D reconstruction results.

[0163] In an alternative embodiment, inputting the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data includes:

[0164] Performing spatial segmentation on the fused point cloud data based on the Euclidean distance between points and the point cloud density to obtain a plurality of point cloud subsets, constructing a KD tree index structure for each point cloud subset, and calculating the nearest neighbor point set of each point based on the KD tree index structure;

[0165] For each nearest neighbor point set, calculating a local structure tensor using a preset tensor voting algorithm, decomposing the local structure tensor based on the eigenvalues to obtain the normal vector and tangent plane of each point; constructing an orthogonal basis function family on the tangent plane, fitting the local surface using the orthogonal basis function family, and calculating the first principal curvature and the second principal curvature of each point according to the local surface;

[0166] Calculate the variances of the first principal curvature and the second principal curvature of each point in the set of neighboring points. When the variance is greater than a preset threshold, mark the corresponding point as a candidate outlier; calculate the spatial distribution density of the points in the set of neighboring points, and verify the candidate outliers based on the spatial distribution density to obtain a set of outliers;

[0167] For the points that do not belong to the set of outliers, construct a Gaussian curvature map based on the first principal curvature and the second principal curvature, calculate the local entropy value of the Gaussian curvature map, determine the local entropy value as the local feature intensity, calculate the shape parameter and the scale parameter of the smoothing kernel function according to the local feature intensity, and construct an adaptive smoothing kernel function; apply the adaptive smoothing kernel function to the set of neighboring points, and calculate the contribution weight of each neighboring point based on the corresponding angle between the Euclidean distance between points and the normal vector. The contribution weight decreases as the Euclidean distance between points increases and decreases as the corresponding angle of the normal vector increases;

[0168] Construct a geometric feature preservation function, which includes a position constraint term and a normal vector consistency constraint term. The position constraint term calculates the weighted average deviation of the point position based on the contribution weight, and the normal vector consistency constraint term calculates the weighted average deviation of the normal vector based on the contribution weight; update the position and normal vector of the point by iteratively optimizing the geometric feature preservation function, and at the same time remove the points in the set of outliers to obtain optimized point cloud data.

[0169] The smoothing kernel function specifically refers to assigning different weights to data points within a certain range, such that points closer to the center have a greater influence, while points farther from the center have a smaller influence; in point cloud data processing, the smoothing kernel function is used to optimize the distribution of points, reduce the influence of noise, and at the same time try to retain the details of the surface structure. It adjusts the contributions of each point in the point cloud to make the overall shape smoother and continuously optimizes the data during the iteration process to make it more conform to the real geometric features; a good smoothing kernel function needs to balance noise reduction and feature preservation. If the kernel function is too smooth, it may lead to the loss of edges and details, making the geometric features blurred; if the smoothing effect is too weak, it may not be able to effectively remove noise. To achieve a balance between smoothing and feature preservation, the influence degree of the kernel function is usually adjusted by combining the spatial distance between points and the directional change of the normal vector; in addition, the smoothing kernel function can adaptively adjust its parameters to adapt to the characteristics of different regions. For example, in flat regions, a stronger smoothing effect can be used, while in regions with larger curvatures, such as corners and boundaries, the smoothing degree is reduced to retain more details. This adaptive adjustment method can improve the accuracy of point cloud optimization and make the final three-dimensional morphology more real and reliable.

[0170] In one specific implementation, the fused point cloud data is first spatially segmented. The Euclidean distances between all point pairs in the point cloud are calculated, and the number of points in each point's neighborhood is counted as the point cloud density. Combining the inter-point distance and density information, the point cloud is divided into multiple small point cloud subsets. To accelerate subsequent processing, a KD tree index structure is constructed for each point cloud subset to quickly search for each point's neighbors.

[0171] Next, local surfaces are fitted for the points in each point cloud subset. Using a KD tree index structure, the K nearest neighbors of each point are found. For each set of neighboring points, a local structure tensor is calculated using a tensor voting algorithm. By performing eigenvalue decomposition on the local structure tensor, the normal vector and tangent plane for each point are obtained. A set of orthogonal basis functions, such as polynomial basis functions, are constructed on the tangent plane and used to fit the local surface. Based on the fitted local surface, the two principal curvatures of each point are calculated: the first principal curvature and the second principal curvature.

[0172] Next, outlier detection is performed. The variance of the two principal curvatures of all points in the neighboring point set is calculated. If the variance exceeds a preset threshold, the corresponding point is marked as a candidate outlier. To avoid misjudgment, the spatial distribution density of the points in the neighboring point set is further calculated. If the candidate outlier is located in a low-density area, it is confirmed as an outlier and added to the outlier set.

[0173] For non-outlier points, local curvature adaptive smoothing is performed. Based on the two principal curvatures of each point, a Gaussian curvature map is constructed. The local entropy value of the Gaussian curvature map is calculated and defined as the local feature strength. Based on the local feature strength, the shape parameter and scale parameter of the smoothing kernel function are calculated. The higher the local feature strength, the more drastic the local curvature change. In this case, the shape parameter and scale parameter of the smoothing kernel function should be smaller to preserve the detailed features. Conversely, the lower the local feature strength, the gentler the local curvature change. In this case, the shape parameter and scale parameter of the smoothing kernel function should be larger to achieve a stronger smoothing effect.

[0174] After constructing an adaptive smoothing kernel function, apply it to a set of neighboring points. Calculate the contribution weight of each neighboring point. The contribution weight is determined by two factors: the Euclidean distance between the points and the angle between their normal vectors. The greater the distance and the larger the angle, the smaller the contribution weight.

[0175] Finally, the point positions and normals are updated by iteratively optimizing a geometric feature-preserving function. This function contains a position constraint and a normal consistency constraint. The position constraint calculates the weighted average deviation of the point positions based on the contribution weights, while the normal consistency constraint calculates the weighted average deviation of the normals based on the contribution weights. By minimizing the geometric feature-preserving function, the point positions and normals are smoothed, while outlier points are removed, ultimately resulting in optimized point cloud data.

[0176] For example, for a point cloud subset containing 1000 points, set the value of K to 20. Calculate the 20 nearest neighbor points of each point, calculate the local structure tensor using the tensor voting algorithm, and fit the local surface. Assume that the two principal curvatures of one point are 0.1 and 0.05 respectively, and the variance of the two principal curvatures of its 20 nearest neighbor points is 0.001, which is less than the preset threshold of 0.01. Then this point is not a candidate outlier. Calculate the local entropy value of the Gaussian curvature map in the region where this point is located to be 0.5, and determine the shape parameter of the smoothing kernel function to be 2 and the scale parameter to be 0.1 according to the local entropy value. Calculate the distance and normal vector angle between this point and its 20 nearest neighbor points to obtain the contribution weight of each nearest neighbor point. Update the position and normal vector of this point by iteratively optimizing the geometric feature preservation function.

[0177] In this embodiment, by removing outliers and smoothing noise, the quality of the point cloud data can be effectively improved, providing a reliable data basis for subsequent applications such as 3D reconstruction and model analysis; adopting a smoothing kernel function with local curvature adaptivity can adjust the smoothing intensity according to the local feature strength, while removing noise, maximizing the retention of the geometric features of the point cloud and avoiding detail loss caused by over-smoothing; combining point cloud density and local curvature information for outlier detection can effectively identify various types of outliers and has strong adaptability to changes in the noise level, ensuring the robustness of the algorithm.

[0178] In an alternative embodiment, the surface of the flexible circuit board is divided into multiple feature sub-regions according to the deformation-sensitive region, and an adaptive spline surface is used to perform block fitting on the feature sub-regions to determine the fitting result; based on the optimized point cloud data, calculate the curvature distribution of the flexible circuit board, and optimize and correct the fitting result in combination with the strain distribution map to generate the three-dimensional geometric morphology of the flexible circuit board, including:

[0179] Obtain the point cloud data on the surface of the flexible circuit board and the position data of the deformation-sensitive region on the surface of the flexible circuit board; extract a set of feature-enhanced points based on the position data of the deformation-sensitive region, calculate the feature importance value of each point in the set of feature-enhanced points, and determine the mesh division parameter according to the feature importance value. The mesh division parameter decreases as the feature importance value increases;

[0180] Use the mesh division parameter to perform adaptive quadtree partitioning on the point cloud data to obtain multiple feature sub-regions; establish a local coordinate system in each feature sub-region, project the point cloud data in the feature sub-region onto the local coordinate system, and calculate the local curvature distribution of the projected point cloud data in the feature sub-region;

[0181] Construct a weight coefficient according to the local curvature distribution, where the weight coefficient increases as the local curvature distribution increases; construct an adaptive spline surface model based on the weight coefficient, and perform surface fitting on the feature sub-region through the B-spline basis function to obtain the initial fitting surface of the feature sub-region;

[0182] Obtain the strain distribution data of the flexible circuit board, calculate the Gaussian curvature value and the mean curvature value of the feature sub-region based on the initial fitting surface, and construct a surface correction function according to the strain distribution data, the Gaussian curvature value and the mean curvature value. The correction amount of the surface correction function is positively correlated with the strain distribution data;

[0183] Calculate the normal vector of the feature sub-region, apply the correction amount of the surface correction function along the direction of the normal vector to the initial fitting surface to obtain an optimized fitting surface; iteratively optimize the parameters of the surface correction function to make the optimized fitting surface match the strain distribution data, and generate the three-dimensional geometric morphology of the flexible circuit board.

[0184] In a specific embodiment, obtain the point cloud data on the surface of the flexible circuit board and the position data of the deformation-sensitive region. The point cloud data can be obtained by a 3D scanner or other depth cameras and contains the three-dimensional coordinate information of each point on the surface of the flexible circuit board. The position data of the deformation-sensitive region can be obtained by manual marking or image processing algorithms. For example, edge detection algorithms can identify the edges of components on the circuit board, and these edges are usually deformation-sensitive regions. Assume that the obtained point cloud data contains 10,000 points, and each point contains three coordinate values of x, y, and z. The position data of the deformation-sensitive region contains 100 points, and each point also contains three coordinate values of x, y, and z.

[0185] Extract a feature-enhanced point set based on the position data of the deformation-sensitive region, and calculate the feature importance value of each point. The feature-enhanced point set can be the point set of the deformation-sensitive region itself or an extended point set generated according to the position data of the deformation-sensitive region. The calculation of the feature importance value can consider the distance from the point to the deformation-sensitive region. The closer the distance, the higher the importance value. For example, a Gaussian function can be used to calculate the feature importance value. The closer the distance to the deformation-sensitive region, the larger the Gaussian function value, that is, the higher the feature importance value. Assume that the feature-enhanced point set contains 200 points, and the feature importance value of each point is between 0 and 1.

[0186] Determine the mesh division parameters according to the feature importance value. The mesh division parameters decrease as the feature importance value increases. This means that in the area with a high feature importance value, the mesh division is finer, while in the area with a low feature importance value, the mesh division is coarser. For example, a linear function can be used to define the relationship between the mesh division parameters and the feature importance value. Assume that when the feature importance value is 1, the mesh division parameter is 0.1, and when the feature importance value is 0, the mesh division parameter is 1.

[0187] Use the mesh division parameters to perform adaptive quadtree partitioning on the point cloud data to obtain multiple feature sub-regions. Adaptive quadtree partitioning is a recursive mesh division method that divides the point cloud data into multiple sub-regions according to the mesh division parameters. In the area with a high feature importance value, the mesh division is finer, and more and smaller sub-regions will be generated. Assume that the point cloud data is divided into 100 feature sub-regions.

[0188] Establish a local coordinate system within each feature sub-region and project the point cloud data within the feature sub-region onto the local coordinate system. The establishment of the local coordinate system can be based on the distribution of the point cloud data within the feature sub-region. For example, the center point of the feature sub-region can be used as the origin of the local coordinate system. Assume that each feature sub-region contains 100 points.

[0189] Calculate the local curvature distribution of the projected point cloud data within the feature sub-region. The local curvature distribution can reflect the degree of curvature of the point cloud data within the feature sub-region. For example, the principal curvature can be used to represent the local curvature distribution.

[0190] Construct the weight coefficient according to the local curvature distribution. The weight coefficient increases as the local curvature distribution increases. This means that in the area with a larger curvature, the weight coefficient is larger, while in the area with a smaller curvature, the weight coefficient is smaller. For example, a linear function can be used to define the relationship between the weight coefficient and the local curvature.

[0191] Construct an adaptive spline surface model based on the weight coefficient and perform surface fitting on the feature sub-region through the B-spline basis function to obtain the initial fitting surface of the feature sub-region. The B-spline basis function is a commonly used surface fitting method that can perform smooth surface fitting on the feature sub-region according to the weight coefficient.

[0192] When performing adaptive spline surface block fitting on the feature sub-region, data preparation for the feature sub-region is carried out. The point cloud data set within each feature sub-region is obtained, a local coordinate system is established for the point cloud data set, and all point cloud data is projected into this coordinate system. By analyzing the projected point cloud data, the local curvature values of all points within the sub-region are calculated, which will be an important basis for subsequent weight assignment. For example, when a feature sub-region containing 100 points is obtained, first, the precise three-dimensional coordinate data of these 100 points is obtained. Then, a local coordinate system is established with the geometric center point of this region as the origin, the data of these 100 points is projected into this coordinate system, and the local curvature value of each point is calculated. These curvature values usually range from 0 to 1.

[0193] After completing the data preparation, it enters the weight coefficient construction stage. The local curvature magnitudes of each point are analyzed in detail, and corresponding weight values are assigned to each point based on the curvature values, following the principle that the greater the curvature, the greater the weight. Finally, all weight values are normalized to ensure numerical stability. In specific implementation, different weight levels can be set according to the curvature values of the points. For example, points with a curvature value above 0.8 are assigned a weight of 1.0, points with a curvature value between 0.5 and 0.8 are assigned a weight of 0.8, points with a curvature value between 0.2 and 0.5 are assigned a weight of 0.5, and points with a curvature value below 0.2 are assigned a weight of 0.3. The differential weight assignment can better maintain the geometric features of the feature region.

[0194] In the B-spline surface fitting stage, the control point grid size of the B-spline surface needs to be determined. Control points are evenly arranged within the sub-region, the corresponding parametric coordinates are calculated for each data point, and a B-spline basis function system is established. Based on this, a weighted least squares equation system is constructed and solved to obtain the optimal positions of the control points, and finally an initial fitting surface is generated. For example, a 6×6 control point grid can be set, 36 control points are evenly distributed within the sub-region, then the parametric coordinates of 100 data points are calculated respectively, a 3rd-order B-spline basis function is established, and the optimal positions of 36 control points are obtained by solving the weighted equation system, and finally a smooth initial fitting surface is generated.

[0195] After obtaining the initial fitting surface, strain data fusion optimization is carried out. The strain distribution data corresponding to the sub-region is obtained, the geometric features of the initial fitting surface are calculated, including Gaussian curvature and mean curvature, and then a surface correction function considering strain is constructed. The normal vectors of each point on the surface are calculated, and the correction amount is applied along the normal vector direction to continuously update the shape of the fitting surface. Specifically, correction rules can be formulated: when the strain value increases by 0.1 each time, the corresponding correction amount increases by 0.05 mm. In this way, a larger correction will be applied in the high-strain region to obtain a more accurate optimized surface.

[0196] When performing iterative optimization, a surface fitting optimization function is set. By continuously adjusting parameters, the matching degree between the fitted surface and the strain data is evaluated until the final fitting result is output after meeting the convergence conditions. Specifically, when performing iterative optimization of the fitted surface, first, the surface fitting optimization objective is defined as the minimum value of the difference between the fitted surface and the actual strain data, which is represented by the error between the predicted value and the actual value at each measurement point. Higher weight values can be assigned to the measurement points in the important areas.

[0197] Initialize the adjustable parameters in the surface correction function. These parameters are used to determine the correction method of the initial fitted surface and are usually set as decimals in the order of one percent, starting from small corrections and gradually adjusting to the optimal state.

[0198] At the start of the optimization phase, calculate the fitting effect under the current parameter settings, check the difference between the fitted values and the actual values at all sampling points, and obtain the overall error value by weighted summation of the differences. Analyze the sensitivity of the current error to each parameter, judge the impact of parameter changes on the error, and accordingly determine the parameter adjustment direction.

[0199] Modify the parameters according to the determined direction. The adjustment amplitude needs to be appropriate to avoid instability or excessive slowness in the optimization process. Set the optimization termination conditions including: the overall error is less than the preset standard, the parameter adjustment amount is extremely small, or the maximum number of optimization times is reached.

[0200] Repeat the steps of evaluating the effect, determining the adjustment direction, and performing the adjustment steps until the termination conditions are met. Check the effect after each round of optimization. When the effect deteriorates, adjust the optimization strategy. When there are optimization fluctuations, reduce the adjustment amplitude; when the progress is slow, increase the adjustment amplitude; when there are local optimization dilemmas, reset the initial values of the parameters. At the same time, focus on the fitting effect in the important areas.

[0201] After the optimization is completed, compare the improvement degree of the error before and after optimization, check the smoothness of the fitted surface, verify the fitting accuracy in the key areas, and ensure that the results have physical significance. Apply the final parameters to the correction function to generate the final fitted surface, and record the key parameter settings and effect data during the optimization process.

[0202] This parameter optimization method improves the matching degree between the fitted surface and the actual strain data through systematic adjustment, ensures the optimization convergence and result accuracy, and can obtain accurate three-dimensional geometric shape reconstruction results.

[0203] Obtain the strain distribution data of the flexible circuit board. The strain distribution data can be obtained through strain sensors or finite element analysis software. Assume that the strain distribution data is a two-dimensional array representing the strain values of each point on the surface of the flexible circuit board.

[0204] Calculate the Gaussian curvature value and the mean curvature value of the characteristic sub-region based on the initial fitting surface. Gaussian curvature and mean curvature are important indicators of surface geometry and can reflect the degree of surface curvature.

[0205] Construct a surface correction function according to the strain distribution data, the Gaussian curvature value and the mean curvature value. The correction amount of the surface correction function is positively correlated with the strain distribution data. This means that in regions with larger strain, the surface correction amount is larger, while in regions with smaller strain, the surface correction amount is smaller.

[0206] Calculate the normal vector of the characteristic sub-region, and apply the correction amount of the surface correction function along the direction of the normal vector to the initial fitting surface to obtain an optimized fitting surface. The normal vector is a vector perpendicular to the surface and can indicate the direction of the surface.

[0207] Generate the three-dimensional geometric morphology of the flexible circuit board by iteratively optimizing the parameters of the surface correction function to match the optimized fitting surface with the strain distribution data. The iterative optimization process can be achieved by minimizing the error between the optimized fitting surface and the strain distribution data.

[0208] In Figure 4 the line graph compares in detail the convergence performance of four iterative optimization methods during the surface fitting process. This technical solution (triangle marker) shows a rapid error decrease in the initial stage of iteration, dropping from 0.165 mm at the 0th iteration to 0.0265 mm at the 5th iteration, a decrease of about 84%; by the 14th iteration, the error has dropped to 0.0042 mm, reaching below the convergence threshold (0.005 mm), and a total of 14 iterations are required, with each iteration taking 28.5 ms. The gradient descent method (square marker) converges slowly and does not reach the convergence threshold until the 16th iteration. The convergence speed is about 14% slower than this solution. Although each iteration only takes 19.2 ms, the total calculation time is still higher than this solution. The Newton method (round marker) performs well in the first 4 iterations, quickly reducing the error to 0.0242 mm, but then the convergence speed slows down significantly, stabilizing at around 0.007 mm and unable to reach the convergence threshold; and each iteration takes 42.7 ms, with the highest calculation cost. The fixed step size method (diamond marker) performs the worst. Even after 20 iterations, the error is still 0.0201 mm, far higher than the convergence threshold. From the perspective of the change in the optimization step size, this technical solution adopts an adaptive step size strategy, with an initial step size of 0.085, which gradually decreases during the iterative process, maintaining fast and stable convergence; the step size of the Newton method changes violently, quickly decreasing from 0.163 to nearly 0, resulting in stagnation of later optimization; the fixed step size method always remains unchanged at 0.01, lacking adaptability. These data fully demonstrate the superiority of this technical solution in the iterative optimization process. Through the strategy of "iteratively optimizing the parameters of the surface correction function", it successfully balances the convergence speed and the calculation efficiency, achieving an ideal optimization effect with fewer iterations.

[0209] In this embodiment, through adaptive spline surface fitting and surface correction, the complex shape of the flexible circuit board can be more accurately described, especially in the deformation-sensitive areas; the adaptive quadtree partitioning can block the point cloud data according to the feature importance value, improving the calculation efficiency; by combining the strain distribution data for surface correction, the influence of noise and errors can be effectively eliminated, improving the reliability of the reconstruction result.

[0210] Figure 5 It is a schematic structural diagram of the three-dimensional topography laser scanning measurement and reconstruction system for the flexible circuit board according to the embodiment of the present invention, as Figure 5 shown, the system includes:

[0211] A first unit for applying a preset stress field to the flexible circuit board to be measured, constructing a grid-shaped strain monitoring point array on the surface of the flexible circuit board, obtaining the strain distribution map of the flexible circuit board through a multispectral camera, and dividing the deformation-sensitive areas based on the strain distribution map; calculating the optimal arrangement positions of four groups of binocular laser scanning devices based on the distribution of the deformation-sensitive areas, and collecting the orthogonal grid projection image group of the flexible circuit board by using an adaptive exposure control strategy;

[0212] A second unit for inputting the calibration data of the four groups of binocular laser scanning devices into an adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix; performing double Gaussian fitting on the orthogonal grid projection image group, extracting the center line of the laser stripe, combining the center line of the laser stripe with the spatial mapping relationship matrix to construct a depth information mapping model, calculating three-dimensional coordinates according to the depth information mapping model, generating initial three-dimensional point cloud data, and performing resampling and registration fusion on the initial three-dimensional point cloud data through a local feature preservation algorithm to obtain the fused point cloud data;

[0213] A third unit for inputting the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data; dividing the surface of the flexible circuit board into multiple feature sub-regions according to the deformation-sensitive areas, performing block fitting on the feature sub-regions by using an adaptive spline surface, and determining the fitting result; calculating the curvature distribution of the flexible circuit board based on the optimized point cloud data, and optimizing and correcting the fitting result in combination with the strain distribution map to generate the three-dimensional geometric topography of the flexible circuit board.

[0214] In the third aspect of the embodiment of the present invention,

[0215] A kind of electronic device is provided, including:

[0216] A processor;

[0217] A memory for storing processor-executable instructions;

[0218] Wherein, the processor is configured to call the instructions stored in the memory to execute the method described above.

[0219] In the fourth aspect of the embodiments of the present invention,

[0220] A computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method described above is implemented.

[0221] The present invention can be a method, an apparatus, a system, and / or a computer program product. The computer program product may include a computer-readable storage medium on which computer-readable program instructions for executing various aspects of the present invention are loaded.

[0222] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for laser scanning measurement and reconstruction of the three-dimensional topography of a flexible circuit board, characterized in that, Including: Applying a preset stress field to the flexible circuit board to be measured, constructing a grid-shaped strain monitoring point array on the surface of the flexible circuit board, obtaining the strain distribution map of the flexible circuit board through a multi-spectral camera, and dividing the deformation-sensitive area based on the strain distribution map; Calculating the optimal arrangement positions of four groups of binocular laser scanning devices based on the distribution of the deformation-sensitive areas, and collecting the orthogonal grid projection image group of the flexible circuit board by using an adaptive exposure control strategy; Inputting the calibration data of the four groups of binocular laser scanning devices into an adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix; performing double Gaussian fitting on the orthogonal grid projection image group, extracting the center line of the laser stripe, combining the center line of the laser stripe with the spatial mapping relationship matrix to construct a depth information mapping model, calculating three-dimensional coordinates according to the depth information mapping model, generating initial three-dimensional point cloud data, and performing resampling and registration fusion on the initial three-dimensional point cloud data through a local feature preservation algorithm to obtain the fused point cloud data; Inputting the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data; dividing the surface of the flexible circuit board into multiple feature sub-regions according to the deformation-sensitive areas, and performing piecewise fitting on the feature sub-regions by using an adaptive spline surface to determine the fitting result; Calculating the curvature distribution of the flexible circuit board based on the optimized point cloud data, and optimizing and correcting the fitting result in combination with the strain distribution map to generate the three-dimensional geometric shape of the flexible circuit board.

2. The method according to claim 1, characterized in that Applying a preset stress field to the flexible circuit board to be measured, constructing a grid-shaped strain monitoring point array on the surface of the flexible circuit board, and dividing the deformation-sensitive area based on the strain distribution map obtained through a multi-spectral camera includes: Applying a preset stress field to the flexible circuit board, constructing a grid-shaped strain monitoring point array on the surface of the flexible circuit board, and obtaining a stressed flexible circuit board with the grid-shaped strain monitoring point array; collecting multi-spectral images of the stressed flexible circuit board before and after applying the preset stress field, and performing registration processing on the multi-spectral images to obtain a registered deformation image; Dividing the registered deformation image into multiple sub-regions, each of the sub-regions containing at least one monitoring point in the grid-shaped strain monitoring point array; constructing a displacement correlation function between adjacent sub-regions, the displacement correlation function including the gray value distribution and displacement increment of the sub-regions; solving the displacement correlation function through an iterative optimization algorithm to obtain the displacement vector of the sub-regions, calculating the displacement gradient of each sub-region based on the displacement vector, and obtaining the displacement field distribution of the flexible circuit board; Interpolate the displacement field distribution using a preset radial basis function to obtain the continuous displacement field of the flexible circuit board. Calculate the strain tensor of the flexible circuit board based on the continuous displacement field, obtain the strain distribution map of the flexible circuit board, calculate the principal strain amplitude according to the strain distribution map, determine the area where the principal strain amplitude exceeds the preset strain threshold as the candidate sensitive area, calculate the strain gradient value within the candidate sensitive area, and obtain the strain gradient value distribution map; Based on the strain gradient value distribution map, determine the points with the maximum strain gradient within the candidate sensitive area as the seed points, and construct a strain monitoring network centered on the seed points. The strain monitoring network includes multiple concentric annular monitoring areas. Determine the radius difference between adjacent concentric annular monitoring areas according to the strain gradient value distribution map, and set multiple strain monitoring nodes within each concentric annular monitoring area; Calculate the correlation degree between the strain value of each strain monitoring node and the strain values of the strain monitoring nodes in the adjacent concentric annular monitoring areas, and determine the strain weight value of each strain monitoring node based on the correlation degree; determine the area corresponding to the strain monitoring node with the strain weight value greater than the preset weight threshold as the deformation sensitive area of the flexible circuit board.

3. The method according to claim 1, characterized in that, Based on the distribution of the deformation sensitive area, calculate the optimal arrangement positions of four groups of binocular laser scanning devices, including: Represent the deformation sensitive area as multiple planar sub-areas, each planar sub-area containing a central coordinate and an area weight. Based on the central coordinate and the area weight, construct a set of position parameters for the four groups of binocular laser scanning devices; Calculate the observation angle of each group of binocular laser scanning devices for each planar sub-area, and at the same time evaluate the laser stripe imaging quality coefficient at each observation angle, and calculate the occlusion coefficient of each planar sub-area at each observation angle; determine the visibility score of the planar sub-area based on the product of the observation angle, the imaging quality coefficient, and the occlusion coefficient, and construct an optimization objective function based on the area weight and the visibility score; Construct a first candidate solution group and a second candidate solution group based on the set of position parameters, each candidate solution carrying a set of the position parameters; the first search radius corresponding to the first candidate solution group is greater than the second search radius corresponding to the second candidate solution group; calculate the optimization objective function value corresponding to each candidate solution; select the candidate solution with the highest optimization objective function value in the first candidate solution group as the first optimal solution, and select the candidate solution with the highest optimization objective function value in the second candidate solution group as the second optimal solution; Centered around the first optimal solution, within the range of the first search radius, a corresponding number of new first position parameter sets are randomly generated according to a preset first quantity, and the candidate solutions carrying the new first position parameter sets are updated to the first candidate solution group to obtain an updated first candidate solution group; centered around the second optimal solution, within the range of the second search radius, a corresponding number of new second position parameter sets are randomly generated according to a preset second quantity, and the candidate solutions carrying the new second position parameter sets are updated to the second candidate solution group to obtain an updated second candidate solution group; Calculate the optimized objective function values of each candidate solution in the updated first candidate solution group and the updated second candidate solution group; if the optimized objective function value of the first optimal solution is greater than that of the second optimal solution, then replace the candidate solution with the lowest optimized objective function value in the second candidate solution group with the position parameter set of the first optimal solution; Repeat the iterative process. Based on a preset number of consecutive iterations, when the maximum improvement amplitude of the optimized objective function value is less than a preset improvement threshold, determine the position parameter set corresponding to the first optimal solution as the optimal layout scheme of the four groups of binocular laser scanning devices.

4. The method according to claim 1, characterized in that, Input the calibration data of the four groups of binocular laser scanning devices into the adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix, including: Obtain the calibration images and calibration data of the binocular laser scanning devices, extract the initial internal parameter values of the binocular laser scanning devices based on the calibration data, and the initial internal parameter values include focal length parameter, principal point coordinate parameter, and radial distortion coefficient parameter; construct a camera internal parameter matrix based on the focal length parameter and the principal point coordinate parameter, and construct a distortion coefficient matrix based on the radial distortion coefficient parameter; combine the camera internal parameter matrix and the distortion coefficient matrix to construct an initial projection matrix; Extract a set of projection points from the calibration images, divide the projection points in the set of projection points into a central region projection point subset and an edge region projection point subset according to the distance to the image center point; apply the initial projection matrix to the central region projection point subset and the edge region projection point subset respectively, and calculate the mean and standard deviation of the reprojection errors of each region; Determine the region weight term according to the mean and standard deviation of the reprojection errors corresponding to the region where the projection point is located, determine the distance weight term according to the distance of the projection point to the image center point, and determine the double adaptive weight factor based on the region weight term and the distance weight term; Multiply the double adaptive weight factor by the square of the reprojection error of the corresponding projection point to construct a weighted reprojection error, use the weighted reprojection error to construct an optimized objective function, and perform optimized calculations on the projection points in the central region projection point subset and the edge region projection point subset respectively based on the optimized objective function to update the initial internal parameter values to obtain an optimized internal parameter matrix; An initial value of the rotation matrix is calculated based on the subset of the projection points in the central region, and the initial value of the rotation matrix is used as the initial value. Combining with the radial distortion characteristics of the subset of the projection points in the edge region, the external parameter matrix is calculated, and the external parameter matrix includes the final rotation matrix parameters and the translation vector parameters; the optimized internal parameter matrix and the external parameter matrix are combined to construct a spatial mapping relationship matrix; The reprojection errors of the projection points in the subset of the projection points in the central region and the subset of the projection points in the edge region are calculated respectively. When the reprojection error of the subset of the projection points in the central region is greater than the first preset threshold or the reprojection error of the subset of the projection points in the edge region is greater than the second preset threshold, the dual adaptive weight factor is updated according to the current reprojection error distribution, and the iteration is continued until the reprojection error of the subset of the projection points in the central region is less than the first preset threshold and the reprojection error of the subset of the projection points in the edge region is less than the second preset threshold. The current spatial mapping relationship matrix is determined as the final spatial mapping relationship matrix.

5. The method according to claim 1, wherein Perform dual Gaussian fitting on the orthogonal grid projection image group, extract the laser stripe center line, combine the laser stripe center line with the spatial mapping relationship matrix to construct a depth information mapping model, calculate the three-dimensional coordinates according to the depth information mapping model, generate the initial three-dimensional point cloud data, and perform resampling and registration fusion on the initial three-dimensional point cloud data through the local feature preservation algorithm. The fused point cloud data obtained includes: The orthogonal grid projection image group is grayscale processed to obtain a grayscale image, and the grayscale image is decomposed into multiple layers of sub-images by a multi-scale decomposition method. The corresponding adaptive window is determined based on the local grayscale gradient amplitude of each layer of sub-images in the multiple layers of sub-images; Sampling points of the laser stripe grayscale distribution are collected in the adaptive window, and the sampling points are subjected to outlier detection to obtain optimized sampling points. A dual Gaussian distribution model is established based on the optimized sampling points, and the dual Gaussian distribution model includes a first Gaussian distribution function and a second Gaussian distribution function; a weighted least squares objective function with a regularization term is constructed, and by minimizing the weighted least squares objective function, the parameters of the first Gaussian distribution function and the second Gaussian distribution function are solved to obtain a multi-scale laser stripe grayscale distribution fitting model; Based on the multi-scale laser stripe grayscale distribution fitting model, the laser stripe center lines of each layer of sub-images in the multiple layers of sub-images are extracted to obtain multi-scale laser stripe center lines. The multi-scale laser stripe center lines are adaptively fused to obtain a laser stripe fused center line. The laser stripe fused center line is combined with the spatial mapping relationship matrix to establish a non-linear spatial mapping relationship, and the spatial coordinate points corresponding to the laser stripe fused center line are calculated according to the non-linear spatial mapping relationship to obtain the initial three-dimensional point cloud data; Calculate the principal curvature parameter and principal direction parameter of each spatial coordinate point in the initial three-dimensional point cloud data, and calculate the point distribution entropy of the spatial coordinate point based on the principal curvature parameter and the principal direction parameter; combine the principal curvature parameter, the principal direction parameter and the point distribution entropy to construct a feature descriptor; calculate the adaptive weight coefficient of the spatial coordinate point based on the feature descriptor; perform hierarchical resampling on the initial three-dimensional point cloud data according to the adaptive weight coefficient to obtain resampled point cloud data; Construct a local consistency constraint including point position constraint, normal vector constraint and curvature constraint, integrate the local consistency constraint into the iterative closest point algorithm to obtain a constraint optimization model; perform iterative registration on the resampled point cloud data based on the constraint optimization model to obtain registered point cloud data; perform local fusion on the registered point cloud data according to the feature descriptor to obtain fused three-dimensional point cloud data.

6. The method according to claim 1, characterized in that, Input the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data, including: Perform spatial segmentation on the fused point cloud data based on the Euclidean distance between points and the point cloud density to obtain multiple point cloud subsets, construct a KD-tree index structure for each point cloud subset, and calculate the nearest neighbor point set of each point based on the KD-tree index structure; For each nearest neighbor point set, use a preset tensor voting algorithm to calculate the local structure tensor, and decompose the local structure tensor based on the eigenvalues to obtain the normal vector and tangent plane of each point; construct an orthogonal basis function family on the tangent plane, fit the local surface using the orthogonal basis function family, and calculate the first principal curvature and the second principal curvature of each point according to the local surface; Calculate the variances of the first principal curvature and the second principal curvature of each point in the nearest neighbor point set, and when the variance is greater than a preset threshold, mark the corresponding point as a candidate outlier; calculate the spatial distribution density of the points in the nearest neighbor point set, and verify the candidate outliers based on the spatial distribution density to obtain an outlier set; For the points that do not belong to the outlier set, construct a Gaussian curvature map based on the first principal curvature and the second principal curvature, calculate the local entropy value of the Gaussian curvature map, determine the local entropy value as the local feature strength, calculate the shape parameter and scale parameter of the smoothing kernel function according to the local feature strength, and construct an adaptive smoothing kernel function; apply the adaptive smoothing kernel function to the nearest neighbor point set, and calculate the contribution weight of each nearest neighbor point based on the Euclidean distance between points and the corresponding angle of the normal vector, and the contribution weight decreases as the Euclidean distance between points increases and decreases as the corresponding angle of the normal vector increases; Construct a geometric feature preservation function, which includes a position constraint term and a normal vector consistency constraint term. The position constraint term calculates the weighted average deviation of the point position based on the contribution weight, and the normal vector consistency constraint term calculates the weighted average deviation of the normal vector based on the contribution weight; update the position and normal vector of the point by iteratively optimizing the geometric feature preservation function, and at the same time remove the points in the outlier set to obtain optimized point cloud data.

7. The method according to claim 1, characterized in that, The surface of the flexible circuit board is divided into multiple characteristic sub-regions according to the deformation-sensitive region, and an adaptive spline surface is used to perform block fitting on the characteristic sub-regions to determine the fitting result; Based on the optimized point cloud data, the curvature distribution of the flexible circuit board is calculated, and the fitting result is optimized and corrected in combination with the strain distribution map to generate the three-dimensional geometric morphology of the flexible circuit board, including: Obtain the point cloud data on the surface of the flexible circuit board and the position data of the deformation-sensitive region on the surface of the flexible circuit board; extract the characteristic enhancement point set based on the position data of the deformation-sensitive region, calculate the characteristic importance value of each point in the characteristic enhancement point set, and determine the mesh division parameter according to the characteristic importance value. The mesh division parameter decreases as the characteristic importance value increases; Use the mesh division parameter to perform adaptive quadtree partitioning on the point cloud data to obtain multiple characteristic sub-regions; establish a local coordinate system in each characteristic sub-region, project the point cloud data in the characteristic sub-region onto the local coordinate system, and calculate the local curvature distribution of the projected point cloud data in the characteristic sub-region; Construct a weight coefficient according to the local curvature distribution, and the weight coefficient increases as the local curvature distribution increases; construct an adaptive spline surface model based on the weight coefficient, and perform surface fitting on the characteristic sub-region through B-spline basis functions to obtain the initial fitting surface of the characteristic sub-region; Obtain the strain distribution data of the flexible circuit board, calculate the Gaussian curvature value and the mean curvature value of the characteristic sub-region based on the initial fitting surface, and construct a surface correction function according to the strain distribution data, the Gaussian curvature value and the mean curvature value. The correction amount of the surface correction function is positively correlated with the strain distribution data; Calculate the normal vector of the characteristic sub-region, apply the correction amount of the surface correction function along the direction of the normal vector to the initial fitting surface to obtain an optimized fitting surface; by iteratively optimizing the parameters of the surface correction function, make the optimized fitting surface match the strain distribution data to generate the three-dimensional geometric morphology of the flexible circuit board.

8. A three-dimensional topography laser scanning measurement and reconstruction system for flexible printed circuit boards, which is used to implement the method described in any one of the preceding claims 1-7, characterized in that, Including: A first unit for applying a preset stress field to the flexible circuit board to be measured, constructing a grid-shaped strain monitoring point array on the surface of the flexible circuit board, obtaining the strain distribution map of the flexible circuit board through a multispectral camera, and dividing the deformation-sensitive region based on the strain distribution map; Based on the distribution of the deformation-sensitive region, calculate the optimal arrangement positions of four groups of binocular laser scanning devices, and collect the orthogonal grid projection image group of the flexible circuit board by using an adaptive exposure control strategy; A second unit, configured to input the calibration data of the four groups of binocular laser scanning devices into an adaptive multi-step calibration algorithm to obtain a spatial mapping relationship matrix; perform double Gaussian fitting on the orthogonal grid projection image group, extract the center line of the laser stripe, combine the center line of the laser stripe with the spatial mapping relationship matrix to construct a depth information mapping model, calculate three-dimensional coordinates according to the depth information mapping model to generate initial three-dimensional point cloud data, and perform resampling and registration fusion on the initial three-dimensional point cloud data through a local feature preservation algorithm to obtain fused point cloud data; A third unit, configured to input the fused point cloud data into a local curvature adaptive algorithm for noise reduction and outlier removal to obtain optimized point cloud data; divide the surface of the flexible circuit board into multiple feature sub-regions according to the deformation sensitive region, and perform block fitting on the feature sub-regions by using an adaptive spline surface to determine the fitting result; calculate the curvature distribution of the flexible circuit board based on the optimized point cloud data, and optimize and correct the fitting result in combination with the strain distribution diagram to generate the three-dimensional geometric shape of the flexible circuit board.

9. An electronic device, characterized in that, Comprising: A processor; A memory for storing instructions executable by the processor; Wherein, the processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, the method according to any one of claims 1 to 7 is implemented.

Images