A copper plate defect detection and inspection method, system, device and medium

Abstract

The present application relates to the technical field of ultrasonic detection, and particularly relates to a copper plate defect detection and inspection method, system, device and medium; the method comprises the following steps: acquiring an original data set; acquiring a point spread function; solving by iterative deconvolution according to the point spread function to acquire a real defect image; extracting the edge of the real defect image to acquire an edge anchor point set; acquiring a spline interpolation function and the like; the interference of the focal spot energy distribution on the defect boundary positioning is eliminated by iterative deconvolution, and the combination of the edge anchor point set and the spline interpolation function ensures the geometric fidelity of the reconstruction model at the edge, while reducing the amount of calculation; the problem that when the focal spot energy loss of the half-wave height method is half of the initial energy, the focal spot center and the edge of the special-shaped defect do not match, resulting in that the measured value is not equal to the actual defect size, is solved; at the same time, the interference of the focal spot energy distribution on the defect boundary positioning is eliminated, and the defect size measurement precision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ultrasonic detection, in particular to a copper plate defect detection and inspection method, system, device and medium. BACKGROUND

[0002] In key fields such as high-end manufacturing, energy and aerospace, the quality and service safety of copper plates and their components are crucial, so accurate non-destructive testing of possible internal volumetric defects (such as corrosion pits, pores, inclusions, etc.) is a core link to ensure their reliability.

[0003] The ultrasonic water immersion focused C-scan technology is the mainstream technology for detecting such defects due to its high resolution and high signal-to-noise ratio. In quantitatively characterizing the size of the defect, the industry usually adopts the half-wave height method, that is, the position where the echo signal amplitude drops to half of the maximum value is used to determine the boundary of the defect.

[0004] Studies have shown that when the ultrasonic focal spot interacts with volumetric defects, due to the uneven distribution of energy inside the focal spot (approximately Gaussian distribution), the focal spot center position that meets the half-wave height condition does not coincide with the physical edge of the defect, which will produce systematic quantitative deviation, usually manifested as the measured value being less than the actual value.

[0005] In addition, after obtaining discrete C-scan data points, an interpolation algorithm is often used to generate a smooth three-dimensional model. Such methods not only have a huge amount of calculation, seriously affecting the detection efficiency, but also their reconstruction results are extremely dependent on the selection of algorithm parameters, and are prone to distortion at the steep edges of the defect. More importantly, the input of such reconstruction methods is the defect boundary and internal data points obtained based on the half-wave height method, which itself has a deviation, and cannot guarantee the geometric fidelity of the final model. SUMMARY

[0006] The main purpose of the present application is to provide a copper plate defect detection and inspection method, which aims to solve the problem that when the half-wave height method loses the focal spot energy to half of the initial energy, the focal spot center does not match the edge of the irregular defect, resulting in a measured value that is not equal to the actual defect size.

[0007] To achieve the above purpose, the present application provides a copper plate defect detection and inspection method, which comprises the following steps:

[0008] Obtain an original data set, the original data set comprising an original C-scan image;

[0009] Obtain a point spread function;

[0010] Solve by iterative deconvolution according to the point spread function to obtain a real defect image;

[0011] extracting edges of the real defect image to obtain an edge anchor set;

[0012] obtaining a spline interpolation function, so that the spline interpolation function satisfies all data points in the edge anchor set;

[0013] creating a three-dimensional grid;

[0014] traversing each point in the three-dimensional grid and determining whether it is inside the contour defined by the edge anchor set;

[0015] If it is outside the contour, the Z coordinate of the point is directly assigned as the original thickness of the copper plate; if it is inside the contour, the coordinate of the point is input into the spline interpolation function to obtain its Z coordinate;

[0016] rendering the point cloud data inside the contour to a detection model with detection information.

[0017] Further, the obtaining of the point spread function specifically includes the following steps:

[0018] manufacturing a point standard defect and performing C scanning to obtain a C scan image, and defining the obtained C scan image as the point spread function.

[0019] Further, the iterative deconvolution solving according to the point spread function includes the following steps:

[0020] iterative deconvolution solving according to the point spread function by Lucy-Richardson deconvolution algorithm.

[0021] Further, the extracting of the edge of the real defect image includes the following steps:

[0022] setting a segmentation threshold;

[0023] obtaining a binary image, and performing threshold segmentation on the binary image according to the segmentation threshold to obtain the binary image;

[0024] extracting the contour of the binary image, and defining the value as the edge anchor set.

[0025] Further, the iterative deconvolution solving according to the point spread function includes the following steps:

[0026] dividing the original C scan image by the convolution of the estimate of the original C scan image and the point spread function to obtain a corrected image;

[0027] spatially inverting the point spread function;

[0028] convolving the corrected image with the spatially inverted point spread function;

[0029] multiplying the convolved image by the estimate of the current C scan image again;

[0030] iterating the above steps to a set number of times to output a real defect image.

[0031] To achieve the above object, the application further provides a copper plate defect detection and inspection system, which comprises:

[0032] a detection module, which is used to acquire an original data set;

[0033] a correction module, which is used to acquire a point spread function; the correction module is further used to solve by iterative deconvolution according to the point spread function to acquire a real defect image; the correction module is further used to extract edges of the real defect image to acquire an edge anchor point set;

[0034] a reconstruction module, which is used to acquire a spline interpolation function so that the spline interpolation function satisfies all data points in the edge anchor point set; the reconstruction module is further used to create a three-dimensional grid; the reconstruction module is further used to traverse each point in the three-dimensional grid and judge whether the point is inside a contour defined by the edge anchor point set; the reconstruction module is further used to render point cloud data inside the contour as a detection model with detection information.

[0035] Further, the detection module comprises a probe and a rotating unit, the probe is arranged at an end of the rotating unit, and the rotating unit is used to adjust different positions of the probe when acquiring the original data set by mechanical rotation.

[0036] Further, the system further comprises an inspection module, which is provided with an ideal model, and the inspection module is used to check whether the detection model is qualified according to the ideal model.

[0037] To achieve the above object, the application further provides a computer device, which comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program.

[0038] To achieve the above object, the application further provides a computer readable storage medium, which stores a computer program, and a processor executes the computer program.

[0039] The application eliminates the interference of focal spot energy distribution on defect boundary positioning by iterative deconvolution, and ensures the geometric fidelity of the reconstructed model at the edge by the combination of the edge anchor point set and the spline interpolation function, while reducing the amount of calculation; when the focal spot energy loss of the half-wave height method is half of the initial energy, the problem that the focal spot center does not match the edge of the special-shaped defect, resulting in that the measured value is not equal to the actual defect size, is solved.

[0040] Meanwhile, the interference of the focal spot energy distribution on the defect boundary positioning is eliminated, the defect size measurement precision is improved, the geometric fidelity of the reconstruction model at the edge is ensured through the combination of the edge anchor point set and the spline interpolation function, the oscillation distortion of the traditional interpolation algorithm at the steep edge of the defect is avoided, only the coordinate calculation is performed on the inside of the defect profile, and the invalid data operation amount is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 A flow chart of the method in the embodiment 1 of the present application is shown in the figure.

[0042] Figure 2 A structure block diagram of the system in the embodiment 9 of the present application is shown in the figure.

[0043] The implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the present application.

[0045] It should be noted that all the directionality indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present application are only used to explain the relative position relationship, movement condition, etc. between the components in a certain posture (as shown in the drawings), and if the certain posture changes, the directionality indications also change accordingly.

[0046] In the present application, unless otherwise explicitly specified and limited, the terms "connection", "fixation" and the like should be understood in a broad sense, for example, "fixation" can be fixed connection, or detachable connection, or integral; can be mechanical connection, or electrical connection; can be direct connection, or indirect connection through an intermediate medium; can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise explicitly limited. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0047] In addition, if the description of "first", "second" and the like is involved in the embodiments of the present application, the description of "first", "second" and the like is only for the purpose of description, and cannot be understood as indicating or implying the relative importance of the technical features indicated or implicitly indicating the number of technical features. Therefore, the features limited by "first", "second" can be explicitly or implicitly included at least one of the features. In addition, the meaning of "and / or" appearing throughout the text includes three parallel schemes. For example, "A and / or B" includes A scheme, or B scheme, or A and B scheme. In addition, the technical solutions of each embodiment can be combined with each other, but it must be based on the realization of ordinary skilled in the art, when the combination of technical solutions appears contradictory or cannot be realized, it should be considered that the combination of technical solutions does not exist, also not within the scope of protection required by the present application.

[0048] Embodiment 1:

[0049] As shown in the accompanying Figure 1 The embodiment provides a copper plate defect detection and inspection method, which comprises the following steps:

[0050] Obtain the original data set, which includes the original C scan image;

[0051] Obtain the point spread function;

[0052] According to the point spread function, iterative deconvolution is solved to obtain the real defect image;

[0053] Extract the edge of the real defect image to obtain the edge anchor point set;

[0054] Obtain the spline interpolation function, so that the spline interpolation function satisfies all data points in the edge anchor point set;

[0055] Create a three-dimensional grid;

[0056] Traverse each point in the three-dimensional grid, and judge whether it is located inside the contour defined by the edge anchor point set;

[0057] If it is outside the contour, the Z coordinate of the point is directly assigned as the original thickness of the copper plate; if it is inside the contour, the coordinate of the point is input into the spline interpolation function to obtain its Z coordinate;

[0058] Render the point cloud data inside the contour to a detection model with detection information.

[0059] It should be noted that in the application of the traditional existing ultrasonic water immersion focusing C-scan technology, the unevenness of the focal spot energy distribution causes systematic deviation of the defect boundary determined by the half-wave height method from the real physical edge, and the interpolation reconstruction method based on discrete data points is difficult to achieve efficient modeling while maintaining geometric fidelity due to the sensitivity of algorithm parameters and computational complexity.

[0060] Based on the above problems, the embodiment provides a copper plate defect detection and inspection method. The original C-scan image is iteratively deconvolved by establishing a point spread function to restore the real defect topography blurred by the focal spot energy distribution. The point spread function is obtained by scanning a point standard defect, which can accurately reflect the spatial response characteristics of the detection system. After obtaining the real defect image, the edge anchor point set is extracted to construct geometric constraints, avoiding the edge distortion caused by the sparse distribution of data points in traditional interpolation algorithms. At the same time, a three-dimensional grid traversal combined with a spline interpolation function is used to calculate the coordinates only inside the defect outline, which not only reduces the amount of invalid data calculation, but also ensures the geometric fidelity of the reconstructed model at the edge.

[0061] It can be understood that the embodiment eliminates the defect size measurement error caused by the unevenness of the ultrasonic focal spot energy distribution by iterative deconvolution, realizes accurate reconstruction of the defect geometric boundary by combining the edge anchor point set and the spline interpolation function, and replaces the traditional unstructured point cloud interpolation with a regularized three-dimensional grid traversal method to improve the calculation efficiency while avoiding model distortion, finally solving the dual technical bottlenecks of systematic deviation of the half-wave height method and insufficient geometric fidelity of the interpolation algorithm.

[0062] It can also be understood that the embodiment obtains an original data set containing an original C-scan image, then obtains a point spread function, which is obtained by scanning a point standard defect and reflects the spatial response characteristics of the detection system, iteratively deconvolves the original C-scan image using the point spread function to solve the real defect image blurred by the focal spot energy distribution, extracts the edge from the real defect image to obtain an edge anchor point set as a geometric constraint, obtains a spline interpolation function that satisfies all data points in the edge anchor point set, creates a three-dimensional grid and traverses each point to determine whether it is inside the outline defined by the edge anchor point set, assigns the Z coordinate of the points outside the outline to the original thickness of the copper plate, and inputs the coordinates of the points inside the outline into the spline interpolation function to obtain the Z coordinate. Finally, the point cloud data inside the outline is rendered into a detection model with detection information. The above method eliminates the interference of focal spot energy distribution on defect boundary positioning by iterative deconvolution, ensures the geometric fidelity of the reconstructed model at the edge by combining the edge anchor point set and the spline interpolation function, reduces the calculation amount, and solves the problem that when the focal spot energy loss is half of the initial energy, the focal spot center does not match the edge of the irregular defect, resulting in a measured value not equal to the actual defect size.

[0063] Meanwhile, by the above method, the embodiment also eliminates the interference of focal spot energy distribution on defect boundary positioning, improves the defect size measurement accuracy, and ensures the geometric fidelity of the reconstructed model at the edge through the combination of the edge anchor point set and the spline interpolation function, avoids the oscillation distortion of the traditional interpolation algorithm at the steep edge of the defect, only calculates the coordinates inside the defect profile, and reduces the invalid data operation amount.

[0064] In some specific embodiments, the copper plate is scanned using an ultrasonic water immersion focusing C-scan system, an original data set containing an original C-scan image is obtained, a spherical standard defect with a diameter of 0.1 mm is manufactured, a point spread function is obtained by C-scan, a Lucy-Richardson deconvolution algorithm is used, the original C-scan image is iteratively deconvolved according to the point spread function, the number of iterations is set to 20 times, a real defect image is obtained, a segmentation threshold is set to 50% of the image gray value, the real defect image is threshold segmented to obtain a binary image, and the profile of the binary image is extracted as an edge anchor point set; a spline interpolation function is used as a spline interpolation function, and all data points in the edge anchor point set are satisfied; a three-dimensional grid with a resolution of 0.01 mm is created; each point in the grid is traversed, and the ray method is used to judge whether the point is located inside the profile defined by the edge anchor point set; for the points outside the profile, the Z coordinate is assigned as the original thickness of the copper plate, 2 mm; for the points inside the profile, the XY coordinates are input into the spline interpolation function to calculate the Z coordinate; finally, the point cloud data inside the profile is rendered into a three-dimensional detection model with depth information using the OpenGL library.

[0065] In some embodiments, the point spread function refers to an image obtained by manufacturing a point-like standard defect and performing C-scan, which can be specifically realized by scanning an artificially prepared point-like defect using an ultrasonic water immersion focusing probe and recording its response signal distribution, and is used to describe the imaging blurring effect of the ideal point defect by the detection system, and eliminate the systematic measurement error caused by uneven focal spot energy distribution.

[0066] In some embodiments, the iterative deconvolution solution refers to the recovery of the real defect image from the blurred original C-scan image by using a mathematical deconvolution algorithm, which can be specifically realized by using the Lucy-Richardson algorithm to update the current estimated image by multiple iterations, and is used to eliminate the influence of uneven ultrasonic focal spot energy distribution on defect boundary positioning and solve the problem of systematic deviation of the half-wave height method in measuring the defect size.

[0067] In some embodiments, the edge anchor point set refers to a set of defect edge coordinates obtained by threshold segmentation and profile extraction, which can be specifically realized by setting a fixed threshold to binarize the real defect image and then extracting the connected region profile point set, and is used to accurately describe the geometric shape of the defect in the two-dimensional plane and avoid the distortion generated by the traditional interpolation method at the steep edge.

[0068] In some embodiments, the spline interpolation function refers to a curved surface function constructed by a mathematical interpolation method.

[0069] In some embodiments, the three-dimensional grid refers to a regularized spatial grid structure composed of discrete points, which can be implemented by assigning a Z coordinate to each grid point after equally dividing the X-Y plane, for efficient generation of a three-dimensional geometric model of the defect, avoiding the problem of low computational efficiency caused by processing unstructured point clouds in traditional interpolation algorithms.

[0070] In some embodiments, the contour inside judgment refers to determining the positional relationship between the grid point and the edge of the defect through geometric calculation, which can be implemented by using the ray method or the polygon inclusion detection algorithm to judge whether the point is located in the closed region defined by the edge anchor point set, for accurately distinguishing the defect area from the normal area and ensuring that the reconstructed model only corrects the thickness inside the defect.

[0071] In the present embodiment, the acquisition of the point spread function specifically includes the following steps:

[0072] A point-like standard defect is manufactured, and C scanning is performed thereon to obtain a C scan image, and the obtained C scan image is defined as a point spread function.

[0073] It can be understood that by pre-preparing a micro-cylindrical hole with a determined geometric parameter on a standard test block as a point-like defect, the processing parameters are strictly controlled to make the defect morphology consistent with the acoustic reflection characteristics of the real volume defect. When performing C scanning, the same detection parameter settings as the copper plate to be detected are adopted to ensure the consistency of the system response. In the two-dimensional image generated by scanning, the gray value of each pixel corresponds to the amplitude attenuation degree of the ultrasonic echo at that spatial position. This data matrix is input into the deconvolution algorithm as the point spread function. By performing deconvolution operation on the C scan data of the actual defect and the point spread function, the edge positioning error caused by uneven distribution of ultrasonic focal spot energy can be effectively eliminated, and the subsequently extracted defect contour is closer to the real physical boundary.

[0074] In the present embodiment, the iterative deconvolution solving according to the point spread function includes the following steps:

[0075] According to the point spread function, the Lucy-Richard deconvolution algorithm is used to perform iterative deconvolution solving.

[0076] It can be understood that the initial estimated image is convolved with the point spread function, and then the original C-scan image is divided by the convolution result pixel by pixel to generate a correction factor matrix. The correction factor matrix is point multiplied with the current estimated image to generate an updated estimated image. After each iteration, the convolution result of the updated estimated image and the point spread function gradually approaches the original C-scan image; when the number of iterations reaches a preset value or the rate of change of the mean square error is lower than a threshold value, the final estimated image is output as the real defect image. Through the algorithm, the blurring effect caused by uneven focal spot energy distribution in the original C-scan image is effectively suppressed, and the geometric features of the defect edge are accurately restored.

[0077] In the embodiment, the edge of the real defect image is extracted by the following steps:

[0078] Setting a segmentation threshold;

[0079] Obtaining a binary image, and performing threshold segmentation on the binary image according to the segmentation threshold to obtain the binary image;

[0080] Extracting the contour of the binary image, and defining the value as an edge anchor point set.

[0081] It should be noted that the segmentation threshold is determined by analyzing the gray scale distribution of the real defect image, for example, the best segmentation threshold is automatically calculated by using the maximum inter-class variance method to eliminate subjective errors caused by manually setting the threshold. During the threshold segmentation process, the gray scale value of each pixel is compared with the segmentation threshold to generate a binary image containing only black and white colors, which effectively distinguishes the defect area from the background area. The contour extraction adopts the Suzuki85 boundary tracking algorithm, which traverses the edge of the white area in the binary image clockwise or counterclockwise, records the coordinate set of all boundary pixel points, and forms a continuous closed contour line. The edge anchor point set obtained through the process has a strict geometric correspondence, which provides accurate boundary definition for the subsequent inside-outside judgment of the three-dimensional mesh, and avoids the model distortion problem caused by the measurement deviation of the half-wave height method.

[0082] In the embodiment, the iteration deconvolution according to the point spread function includes the following steps:

[0083] Dividing the original C-scan image by the convolution of the estimate of the original C-scan image and the point spread function to obtain a correction image;

[0084] Spatially inverting the point spread function;

[0085] Convolving the correction image with the spatially inverted point spread function;

[0086] Multiplying the image after the convolution operation back to the estimate of the current C-scan image;

[0087] Iterating the above steps to a preset number of times to output the real defect image.

[0088] It should be noted that the original C-scan image is divided by the convolution result of the current estimated image and the point spread function to generate a correction image reflecting the error distribution; the flipped point spread function is convolved with the correction image to obtain the inverse propagation amount of the error in the spatial domain; the current estimated image is updated by multiplication operation to form a new iteration input; after the process is repeated for a fixed number of times, the edge positioning deviation caused by uneven distribution of ultrasonic focal spot energy is systematically suppressed, and the edge sharpness of the finally output real defect image is improved, providing accurate geometric boundary data basis for subsequent three-dimensional reconstruction.

[0089] It should also be noted that the detection information includes the real defect image, and also includes the maximum depth, average depth, defect surface area, material loss volume, aspect ratio, etc. of the real defect image.

[0090] Embodiment 2:

[0091] In order to make the technical solutions of the present application more clear, the related content of the original data set is described in detail here. Specifically, the obtained original data set includes a plate surface point set, a defect internal point set and an original C-scan image.

[0092] The specific acquisition steps include:

[0093] Start the three-dimensional motion system controlled by the PLC, carry the high-frequency water immersion point focusing probe, and automatically scan the specified area by C-scan;

[0094] At each scanning coordinate point (x, y), not only the echo amplitude is recorded, but also the complete A-scan (echo) signal waveform is acquired and stored;

[0095] Set an amplitude threshold (for example, points below 90% of the normal surface echo), and preliminarily classify all scanning points to obtain three original data sets;

[0096] Among them, the plate surface point set S: , wherein, respectively represent the length, width and height of the copper plate edge obtained by the A-scan signal.

[0097] Defect internal point set : , wherein, respectively represent the length, width and height of the copper plate edge obtained by the A-scan signal.

[0098] Original C-scan image : a two-dimensional image, and the gray value of each pixel corresponds to the echo energy amplitude of the point (x, y).

[0099] Embodiment 3:

[0100] To make the technical solutions of the present application more clear, the related content of obtaining the point spread function is described in detail, specifically, the point spread function is preferably obtained by the following method:

[0101] Theoretical method: the theoretical calculation is performed through the sound field directivity function (Bessel function), and a Gaussian distribution model of the focal spot energy is fitted, and the two-dimensional Gaussian function is the theoretical point spread function;

[0102] Experimental method: a point-like standard defect (preferably a micron-level pinhole) with a size much smaller than the focal spot diameter of the probe is manufactured, and then C scanning is performed thereon, and the obtained C scan image can be approximately regarded as the most real experimental point spread function of the system.

[0103] It can be understood that in the above two methods, the experimental method is preferred, and no matter which method is used, a two-dimensional function or a matrix representing the blur kernel is obtained.

[0104] Example 4:

[0105] To make the technical solutions of the present application more clear, the related content of the Lucy-Richard deconvolution algorithm is described in detail, and it can be understood that the essence of solving by iterative deconvolution through the Lucy-Richard deconvolution algorithm is to use a known point spread function to restore the most real defect image from the obtained C scan image .

[0106] It can also be understood that the Lucy-Richard deconvolution algorithm is used for solving, and the expression satisfies:

[0107] ;

[0108] wherein, is the estimation of the real defect image at the t+1th iteration;

[0109] is the estimation of the real defect image at the tth iteration, and it can be understood that the initial estimation of the algorithm is the C scan image .

[0110] is the actually observed C scan image with defects, and in the subsequent definition, it is a defect image, and the pixel value thereof represents the echo energy of each point;

[0111] is the point spread function;

[0112] represents the convolution operation;

[0113] Representing the point spread function as spatially flipped, in some embodiments, it is also possible to perform row and column flips in the matrix.

[0114] For the above expression, by convolving the current estimate of the clear image (the estimated image) with the point spread function, this can be thought of as a simulation of the following: if the estimated clear image is correct, then take the image convolved by the system.

[0115] Then, divide the actually observed defective image by the simulated acquired image, to obtain an image with a "correction factor", which is greater than 1 if the simulated blur is not enough at a certain point, or less than 1 if the simulation is too blurred.

[0116] Finally, multiply this "correction factor" image (after a correlation operation) back to the current estimate to obtain a better, new estimate This process is repeated (for example, 5-20 iterations) and each iteration makes the estimated image sharper, with edges closer to the step change, thus approaching the true geometry of the defect.

[0117] It can also be understood that the above algorithm is robust when dealing with noisy images and can well preserve the edge details in the image.

[0118] Embodiment 5:

[0119] In order to make the technical solutions of the present application clearer, the related content of extracting the edge of the real defect image is described in detail, and it can be understood that the purpose of edge extraction is to extract the accurate edge coordinates from the clear defect image obtained after deconvolution and define them as an edge anchor set, which specifically includes the following steps:

[0120] Set a segmentation threshold;

[0121] Obtain a binary image, and perform threshold segmentation on the binary image according to the segmentation threshold to obtain a binary image,

[0122] Extract the contour of the binary image. The contour is a set of two-dimensional coordinate points (edge anchor set).

[0123] Embodiment 6:

[0124] In order to make the technical solutions of the present application clearer, the related content of the spline interpolation function is described in detail, and it can be understood that for each data point in the edge anchor set, Understandably, for all data points, it is necessary to find an optimal smooth surface to fit these points; that is, it is necessary to construct a surface that includes a linear component. (representing the overall tilt trend) and non-linear components ( A smoothing function (representing local fluctuations) is obtained by satisfying this function for all known internal data points (i.e., for each...). They all A system of linear equations can be established to solve for all the unknown coefficients. and Once these coefficients are determined, this function can be used to calculate the depth at any location inside the defect, thereby generating a complete and smooth internal surface.

[0125] It is also understandable that the expression of the spline interpolation function satisfies:

[0126] ;

[0127] in, It is the function to be solved, representing the predicted depth (or remaining thickness) at any coordinate (x, y) inside the defect.

[0128] These are the model coefficients obtained by solving a system of linear equations;

[0129] N is the total number of data points in the defect's internal point set;

[0130] These are the two-dimensional coordinates of the i-th known internal data point;

[0131] It is the two-dimensional coordinate of any point whose depth you want to predict;

[0132] Represents the radial basis functions (kernel functions), where It is a known point With the point to be predicted The Euclidean distance between them.

[0133] Understandably, compared to existing locally weighted linear regression methods, this embodiment only needs to solve a linear system once to determine all coefficients, and then the depth of any point can be quickly calculated, resulting in significantly higher computational efficiency. At the same time, the physical meaning of its minimum bending energy is very consistent with the smooth transition morphology of defects such as corrosion pits and depressions.

[0134] Example 7:

[0135] To make the technical solutions of the present application more clear, the related content of three-dimensional reconstruction is described in detail here, which includes the following steps:

[0136] Creating a high-resolution three-dimensional grid as the carrier of the final model;

[0137] Traverse each point in the grid , specifically including the following steps:

[0138] Determine whether the point is inside the contour defined by the edge anchor set.

[0139] If outside the contour, the Z coordinate of the point is directly assigned as the original thickness of the copper plate ;

[0140] If inside the contour, substitute the coordinates of the point into the spline interpolation function solved in the previous step to calculate its Z coordinate, and the calculation expression satisfies:

[0141] ;

[0142] Through the above steps, a high-precision three-dimensional defect model composed of three parts seamlessly spliced is generated: an accurate flat plate surface, a sharp edge constrained by high-precision data, and a smooth transition defect cavity.

[0143] It can also be understood that the finally generated high-resolution three-dimensional point cloud data can be rendered using tools such as MATLAB to generate intuitive three-dimensional model graphs, and the model can be output for subsequent finite element analysis, volume calculation, or life assessment.

[0144] Example 8:

[0145] In other embodiments, for water immersion inspection of copper plate components with locally shaped surfaces, the interaction between the ultrasonic focal spot and the surface (such as the incident angle, the defocusing amount) may be different at each point, which means that the blur kernel causing image blur is no longer fixed, but changes at different positions of the image. Using a single, fixed point spread for global deconvolution, there is a situation where some areas are undercorrected and some areas are overcorrected. For this problem, the present embodiment provides a differential three-dimensional reconstruction method based on reference surface fitting on the basis of the foregoing embodiments.

[0146] It can be understood that the differential three-dimensional reconstruction method aims to decouple the complex geometric problem into:

[0147] Separation: completely separate the overall reference surface geometry of the copper plate from the local defect geometry, no longer assuming that the plate is flat, but recognizing that it has a complex, curved surface shape that needs to be accurately described.

[0148] Reconstruction: Mathematical modeling and reconstruction are performed on the two independent geometric elements, the reference surface and the relative defect profile.

[0149] Superposition: Finally, through mathematical operations, the reconstructed defect model is precisely superimposed or embedded onto the reconstructed reference surface to obtain the final, high-fidelity three-dimensional model.

[0150] It is also understandable that the method specifically includes the following steps:

[0151] Reference surface fitting: Obtain the set of points on the plate surface (i.e., all scanned points outside the defect profile). Since these points represent the original surface of the plate, a continuous reference surface function is constructed by using thin plate spline interpolation or low-order polynomial surface fitting. This function can be based on any given... The coordinates are used to predict the Z-coordinate height of the defect-free location; the goal of this step is to create an accurate mathematical model to describe the ideal surface of the copper plate in a defect-free state.

[0152] The relative defect depth is obtained from the reference surface function, and the expression satisfies:

[0153] ;

[0154] in, This represents the relative defect depth of the i-th internal point. It is a positive value, representing the depth of material loss at that point.

[0155] Calculated using a reference surface model, in The appropriate, unblemished height at the location.

[0156] This indicates the actual height measured by the ultrasonic wave at that point.

[0157] It should be noted that the essence of the above expression is to perform a mathematical correction to remove the influence of the macroscopic curvature of the board itself on the measurement, and thus separate the geometric changes caused purely by the defect itself.

[0158] Defect contour interpolation: Incorporates all interior points into the input data points. And as a boundary constraint, it is defined as a new set of edge anchor points, for any point on the edge. Its relative depth is defined as And the expression satisfies:

[0159] ;

[0160] in, is the function to be solved, representing the predicted depth (or residual thickness) at any coordinate within the defect;

[0161] are the model coefficients obtained by solving the linear equations;

[0162] M is the total number of data points in the set of points within the defect;

[0163] is the two-dimensional coordinate of the jth data point;

[0164] represents the radial basis function (kernel function), and is the same as the previous embodiments.

[0165] It can be understood that the above expression, similar to the previous embodiments, is to generate a smooth surface, but the key difference is that this embodiment fits the relative depth, i.e. the concave shape of the defect and this shape is precisely constrained within the edge profile with a depth of zero, ensuring a smooth transition of the defect to the reference surface.

[0166] Model synthesis: for any point on the high-resolution three-dimensional grid its final height is obtained by subtracting the relative defect depth of the point from the reference surface height, and the expression satisfies:

[0167] ;

[0168] is the final reconstructed height at coordinate ;

[0169] is the reference surface height of the point;

[0170] is the relative defect depth of the point, in particular, for points outside the profile, the value of is 0.

[0171] Embodiment 9:

[0172] As shown in the accompanying drawings, Figure 2 the present embodiment provides a copper plate defect detection and inspection system, which comprises:

[0173] a detection module, the detection module is used for acquiring an original data set;

[0174] a correction module, the correction module is used for acquiring a point spread function; the correction module is further used for iterative deconvolution solving according to the point spread function to acquire a real defect image; the correction module is further used for extracting an edge of the real defect image to acquire an edge anchor point set;

[0175] a reconstruction module configured to obtain a spline interpolation function, so that the spline interpolation function satisfies all data points in the edge anchor set; the reconstruction module is further configured to create a three-dimensional grid; the reconstruction module is further configured to traverse each point in the three-dimensional grid and determine whether it is located inside the contour defined by the edge anchor set; the reconstruction module is further configured to render the point cloud data located inside the contour as a detection model with detection information.

[0176] It can be understood that the detection module adjusts the position of the probe through the rotating unit to collect original C-scan image data of different regions, and the correction module uses the point spread function to iteratively deconvolve the original data to eliminate edge deviation caused by uneven focal spot energy distribution and extract accurate edge anchor set of the real defect. The reconstruction module generates a spline interpolation function that satisfies all data points based on the edge anchor set, and distinguishes between inside and outside points through a contour judgment mechanism after constructing a three-dimensional grid. The outside points are directly assigned with the original thickness to maintain the background geometric fidelity, and the inside points are calculated for Z coordinates through the spline interpolation function to accurately represent the defect morphology. The final rendered detection model avoids the parameter sensitivity and steep edge distortion problems of traditional interpolation algorithms, and ensures the geometric accuracy of the model based on the accurate edge anchor set.

[0177] In the present embodiment, the detection module includes a probe and a rotating unit, the probe is arranged at the end of the rotating unit, and the rotating unit is used to adjust the different positions of the probe when acquiring the original data set through mechanical rotation.

[0178] It can be understood that the rotating unit controls the movement of the probe according to a preset scanning path, and the probe collects C-scan image data at a set interval angle or displacement during rotation. The position adjustment of the probe is realized through closed-loop control feedback of the encoder inside the rotating unit, ensuring that the position information of each data point corresponds to the actual coordinates of the copper plate surface. The motion trajectory planning of the rotating unit is based on the three-dimensional model of the copper plate, and the entire detection area is gradually covered through segmented rotation. After the probe completes multi-angle data acquisition under the driving of the rotating unit, the system performs spatial registration on the data sets at different positions to form a complete original data set for subsequent deconvolution and defect reconstruction.

[0179] In some embodiments, the probe can be an ultrasonic probe for transmitting and receiving ultrasonic signals.

[0180] In some embodiments, the rotating unit can be a motor-driven rotating arm, and the probe is fixed at the end of the rotating arm. The rotating arm can rotate 360 degrees in the horizontal plane, and can move up and down in the vertical direction.

[0181] In some other embodiments, the rotating unit comprises a rotating inner rod and a rotating outer rod, one end of the rotating outer rod is connected to the movable end of the three-dimensional guide rail of the system through the driving unit, the other end of the rotating outer rod is rotatably arranged with the rotating inner rod through the driving unit, and the probe is arranged on the rotating inner rod through the driving unit. It can be understood that the driving unit is a high-precision motor with a speed reduction structure.

[0182] It can also be understood that the rotating inner rod and the rotating outer rod are quarter fan ring structures with different diameters, and the diameter of the rotating outer rod is greater than that of the rotating inner rod.

[0183] It should be noted that the global linear motion provided by the three-dimensional guide rail and the local angular motion (two-axis rotation) provided by the double-layer fan ring mechanism are combined, and the system can control the multiple degrees of freedom through the PLC control algorithm to make the probe move along an arbitrary complex spatial trajectory, so as to be suitable for local surface water immersion flaw detection. The rotation of the rotating outer rod provides the first dimension of attitude adjustment, and the rotation of the rotating inner rod relative to the outer rod provides the second dimension of attitude adjustment. This decoupled rotation design allows the pointing direction of the probe to be accurately controlled to maintain a good detection surface. In addition, by driving a single fan ring to rotate, the probe can realize a precise circular arc trajectory motion, which is extremely efficient for detecting cylindrical, spherical or cornered components, avoiding the complex calculation and mechanical error of using linear motion to approximate a curve.

[0184] Embodiment 10:

[0185] The system further comprises an inspection module, wherein an ideal model is set in the inspection module, and the inspection module is used to check whether the detection model is qualified according to the ideal model.

[0186] It can be understood that after the detection model is generated, the inspection module loads the reference data of the ideal model, unifies the spatial positions of the detection model and the ideal model through coordinate alignment, calculates the geometric deviation of each point of the detection model from the corresponding position of the ideal model, and if all deviation values are within the allowable error range, the detection model is determined to be qualified; if there is an area exceeding the error range, the defect position is marked and an unqualified conclusion is output. By replacing manual visual inspection with automatic comparison, subjective judgment errors are avoided, and the objectivity of the detection conclusion is ensured through the pre-set quantitative threshold. The data structure of the ideal model adopts the same grid format as the detection model, ensuring the matching of the data dimensions during the comparison process and reducing the consumption of computing resources.

[0187] As a preferred embodiment, the preferred implementation of the embodiment solution is as follows: the inspection module is built-in with a database of ideal three-dimensional models established through finite element simulation, which contains standard geometric parameters corresponding to different defect types; after the detection model is generated, the detection model is spatially aligned with the ideal model corresponding to the defect type through a three-dimensional registration algorithm, and the surface profiles of the two models are compared point by point using a point cloud distance calculation method. If the coordinate deviation of all vertices on the surface of the detection model from the corresponding area of the ideal model is less than the preset threshold value of 0.2 mm, it is determined to be qualified; if more than 10% of the vertices deviate beyond the threshold range, an alarm prompt is triggered and the deviation exceeding the limit area coordinates are automatically labeled, and the inspection result is marked as unqualified.

[0188] Through the above technical solution, the embodiment effectively solves the geometric distortion problem of the traditional detection model caused by boundary positioning deviation, and through the standardized ideal model comparison mechanism, the systematic measurement error caused by uneven focal spot energy distribution can be accurately identified, and the final detection model can truly reflect the physical profile of the defect in the spatial dimension, thereby providing a high-fidelity three-dimensional data basis for subsequent service safety evaluation.

[0189] Embodiment 11:

[0190] In this embodiment, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program.

[0191] The embodiment provides a processor, which is used for running a program, wherein the program performs the copper plate defect detection and inspection method when running.

[0192] The embodiment provides a device, which comprises a processor, a memory, and a program stored on the memory and capable of running on the processor, and the processor performs the program to realize at least the copper plate defect detection and inspection method.

[0193] The device in the present application can be a server, a PC, a PAD, a mobile phone, etc.

[0194] The present application also provides a computer program product adapted to perform a program for initializing at least one copper plate defect detection and inspection method step when executed on a data processing device.

[0195] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0196] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0197] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0198] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0199] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process.Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0200] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0201] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0202] Computer-readable media include permanent and non-permanent, removable and non-removable media. Information storage can be achieved by any method or technology; the information can be computer-readable instructions, data structures, program modules or other data.

[0203] Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random-access memory (SRAM), dynamic random-access memory (DRAM), other types of random-access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media do not include temporary computer-readable media such as modulated data signals and carrier waves.

[0204] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method of copper sheet defect detection and inspection, characterized by, The method comprises the following steps: Obtaining an original data set comprising original C-scan images; Obtaining a point spread function; Solving by iterative deconvolution according to the point spread function to obtain a real defect image; Extracting the edge of the real defect image to obtain an edge anchor point set; Obtaining a spline interpolation function so that the spline interpolation function satisfies all data points in the edge anchor point set; Creating a three-dimensional grid; Traversing each point in the three-dimensional grid and determining whether it is inside the contour defined by the edge anchor point set; If it is outside the contour, the Z coordinate of the point is directly assigned as the original thickness of the copper plate; if it is inside the contour, the coordinates of the point are input into the spline interpolation function to obtain the Z coordinate; Rendering the point cloud data inside the contour into a detection model with detection information. The expression of the spline interpolation function satisfies: ； wherein, is the function to be solved, representing the predicted depth at an arbitrary coordinate (x, y) within the defect; x , y ) within the defect; model coefficients obtained by solving the linear equation system; N Ntotai is the total number of data points in the set of internal points of the defect; is the second known internal data point in two dimensions; and i is the first known internal data point in two dimensions. is the two-dimensional coordinate of any one point that wants to predict the depth; denotes a radial basis function, where is a known point the Euclidean distance between P and the point to be predicted.

2. A method of copper sheet defect detection and inspection as claimed in claim 1, wherein, The obtaining of the point spread function specifically comprises the following steps: Manufacturing a point-like standard defect and C-scanning it to obtain a C-scan image, which is defined as the point spread function.

3. A method of copper sheet defect detection and inspection as claimed in claim 1, wherein, The solving by iterative deconvolution according to the point spread function comprises the following steps: Solving by iterative deconvolution according to the point spread function through the Lucy-Richardson deconvolution algorithm.

4. The method of claim 1, wherein the step of detecting and inspecting the defects of the copper plate is performed by using a camera. The extraction of the edge of the real defect image comprises the following steps: Setting a segmentation threshold; Obtaining a binary image and performing threshold segmentation according to the segmentation threshold to obtain the binary image; Extracting the contour of the binary image and defining the value as the edge anchor point set.

5. The method of claim 1 or 3, wherein the step of detecting and inspecting the defects of the copper plate is performed by using a camera. The solving by iterative deconvolution according to the point spread function comprises the following steps: Dividing the original C-scan image by the convolution of the estimate of the original C-scan image and the point spread function to obtain a corrected image; Spatially inverting the point spread function; Convolving the corrected image with the spatially inverted point spread function; Multiplying the image after the convolution operation by the estimate of the current C-scan image; Iterating the above steps for a set number of times and outputting the real defect image.

6. A copper sheet defect detection and inspection system characterized by, The system comprises: A detection module for obtaining an original data set; A correction module for obtaining a point spread function; the correction module is also used for solving by iterative deconvolution according to the point spread function to obtain a real defect image; the correction module is also used for extracting the edge of the real defect image to obtain an edge anchor point set; A reconstruction module for obtaining a spline interpolation function so that the spline interpolation function satisfies all data points in the edge anchor point set; the reconstruction module is also used for creating a three-dimensional grid; the reconstruction module is also used for traversing each point in the three-dimensional grid and determining whether it is inside the contour defined by the edge anchor point set; the reconstruction module is also used for rendering the point cloud data inside the contour into a detection model with detection information. The expression of the spline interpolation function satisfies: ； wherein, is the function to be solved, representing the predicted depth at an arbitrary coordinate (x, y) within the defect; x , y ) within the defect; model coefficients obtained by solving the linear equation system; N Ntotai is the total number of data points in the set of internal points of the defect; is the two-dimensional coordinate of the nth known internal data point; and i is the two-dimensional coordinate of the nth known internal data point; and is the two-dimensional coordinate of any one point that wants to predict the depth; denotes a radial basis function, where is a known point the Euclidean distance between the point P to be predicted and the known point 7. A copper sheet defect detection and inspection system as claimed in claim 6, wherein, The detection module comprises a probe and a rotating unit, the probe is arranged at the end of the rotating unit, and the rotating unit is used for adjusting different positions of the probe when obtaining an original data set through mechanical rotation.

8. A copper sheet defect detection and inspection system as claimed in claim 6, wherein, The system further comprises an inspection module, an ideal model is arranged in the inspection module, and the inspection module is used for checking whether the detection model is qualified according to the ideal model.

9. A computer device, comprising: The computer device comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the copper plate defect detection and inspection method in any one of claims 1-5.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the processor executes the computer program to realize the copper plate defect detection and inspection method in any one of claims 1-5.

Images