A visual guidance-based semiconductor laser cleaning method

Abstract

A kind of visual guidance-based semiconductor laser cleaning method, it is related to semiconductor component laser cleaning technical field, the three-dimensional geometric data and surface image data of workpiece to be cleaned are collected, and the three-dimensional model of workpiece to be cleaned is constructed;Point cloud analysis is carried out on the three-dimensional model, and the gradient, Gaussian curvature, average curvature and texture feature set of each point in the three-dimensional model are marked as key features;The three-dimensional model is converted into a topological undirected graph, the key points and background points are marked in the topological undirected graph, the weight mark of each undirected edge in the topological undirected graph is set, and the nodes in the topological undirected graph are divided into key points and background points;The path planning is carried out on the key points in the three-dimensional model, the laser cleaning path is obtained, the laser parameter control model is constructed, and the laser parameters of each key point in the laser cleaning path are obtained;The laser cleaning operation of workpiece to be cleaned and laser parameter control model are monitored and feedback corrected in real time, so as to improve cleaning quality and efficiency.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of laser cleaning of semiconductor components, and particularly relates to a semiconductor laser cleaning method based on visual guidance. BACKGROUND

[0002] CN116078751A "Pulse laser cleaning control device and cleaning control method" comprises the following steps: after the host computer module is connected with the control unit module, the host computer module sends control command parameters to the control unit module; the control unit module controls the laser generating module to perform corresponding laser operation by data interaction with the laser generating module based on the received control command parameters.

[0003] CN117798137A "Intelligent laser cleaning control system" comprises the following steps: a collection module collects three-dimensional point cloud data of a steering frame in real time; a processing module generates frame three-dimensional coordinates, deflection angles and surface color data of the steering frame according to the three-dimensional point cloud data, and processes the surface color data to obtain laser output instructions; a transportation module transports the steering frame to a cleaning station according to the frame three-dimensional coordinates, and the cleaning station is provided with a composite laser cleaning device; a position adjustment module adjusts the three-dimensional spatial position of the composite laser cleaning device according to the frame three-dimensional coordinates and the deflection angles; a first emission unit outputs semiconductor laser to the steering frame according to the laser output instructions, so as to reduce the bonding force between the metal material of the steering frame and the surface attachments; and a second emission unit outputs pulse laser to the steering frame according to the laser output instructions, so that the surface attachments are separated from the metal material.

[0004] The prior art usually adopts manual teaching, can only perform simple plane scanning planning, controls a mechanical hand to perform non-discriminatory scanning from left to right and from top to bottom, and can only perform cleaning of the top surface of a workpiece, and cannot clean other surfaces in the case of turning over, so that the limitation is high, and the laser parameters cannot be dynamically adjusted. SUMMARY

[0005] In order to solve the above technical problems, the purpose of the present application is to provide a semiconductor laser cleaning method based on visual guidance, comprising the following steps:

[0006] Step s1: three-dimensional scanning terminal is used to collect three-dimensional geometric data and surface image data of a workpiece to be cleaned, and a three-dimensional model of the workpiece to be cleaned is constructed according to the three-dimensional geometric data;

[0007] Step s2: point cloud analysis is performed on the three-dimensional model to obtain the gradient, Gaussian curvature and average curvature of each point in the three-dimensional model, texture analysis is performed on the surface image data of the workpiece to be cleaned to obtain the texture feature set of each point in the three-dimensional model, and the gradient, Gaussian curvature, average curvature and texture feature set of each point in the three-dimensional model are marked as key features;

[0008] Step s3: converting the three-dimensional model into a topological undirected graph, marking key points and background points in the topological undirected graph, setting weight labels of each undirected edge in the topological undirected graph, and dividing nodes in the topological undirected graph into key points and background points;

[0009] Step s4: path planning for the key points in the three-dimensional model, obtaining a laser cleaning path, constructing a laser parameter control model, and obtaining laser parameters of each key point in the laser cleaning path;

[0010] Step s5: real-time monitoring and feedback correction of the laser cleaning operation of the workpiece to be cleaned and the laser parameter control model.

[0011] Further, the process of obtaining the gradient, Gaussian curvature and average curvature of each point in the three-dimensional model includes:

[0012] Obtaining point cloud data contained in the three-dimensional model, pre-processing the point cloud data, obtaining a normal vector of each point in the pre-processed point cloud data, defining a neighborhood for each point, fitting a local plane in the neighborhood of each point, and obtaining the gradient of each point based on the fitted local plane of each point, and obtaining the Gaussian curvature and average curvature of each point.

[0013] Further, the process of obtaining the texture feature set of each point in the three-dimensional model by performing texture analysis on the surface image data of the workpiece to be cleaned includes:

[0014] Pre-processing the surface image data of the workpiece to be cleaned, obtaining a gray level co-occurrence matrix of each pixel point in the pre-processed surface image data, obtaining a texture feature set of each pixel point according to the gray level co-occurrence matrix of each pixel point, matching each pixel point in the surface image data with each point in the point cloud data of the surface of the three-dimensional model, obtaining a corresponding relationship between each pixel point and each point in the point cloud data, and obtaining a texture feature set of each point in the point cloud data according to the corresponding relationship between each pixel point and each point in the point cloud data.

[0015] Further, the process of converting the three-dimensional model into a topological undirected graph includes:

[0016] Obtaining a key feature corresponding to each point in the point cloud data contained in the three-dimensional model, converting the three-dimensional model into a topological undirected graph, taking each point in the point cloud data contained in the three-dimensional model as a node of the topological undirected graph, and importing the key feature corresponding to each point in the point cloud data contained in the three-dimensional model into the node corresponding to each point;

[0017] Marking the key points and the background points in the topological undirected graph, constructing the undirected edges between the adjacent nodes in the topological undirected graph, and constructing the undirected edges between each node and the key points and between each node and the background points in the topological undirected graph.

[0018] Further, the process of marking the key points and the background points in the topological undirected graph includes:

[0019] Comparing the key features between the adjacent nodes to obtain the key feature similarity between the adjacent nodes, setting the weight label of the undirected edge between the adjacent nodes according to the key feature similarity between the adjacent nodes, and comparing the key features of each node with the key features of the key points and the background points respectively to obtain the key feature similarity between each node and the key points and the key feature similarity between each node and the background points, setting the weight label of the undirected edge between each node and the key points according to the key feature similarity between each node and the key points, and setting the weight label of the undirected edge between each node and the background points according to the key feature similarity between each node and the background points;

[0020] Constructing an energy function according to the weight labels of the undirected edges in the topological undirected graph, segmenting the topological undirected graph based on the energy function using a graph cut algorithm, and dividing the nodes in the topological undirected graph into key points and background points according to the segmentation result.

[0021] Further, the process of marking the key points and the background points in the topological undirected graph includes:

[0022] Obtaining the standard texture parameters of the workpiece to be cleaned, performing feature matching between the texture feature set of each point in the topological undirected graph and the standard texture parameters to obtain the texture parameter similarity between each point and the standard texture parameters, screening out the point corresponding to the minimum texture parameter similarity, marking the point as a key point, and screening out the point corresponding to the maximum texture parameter similarity, marking the point as a background point.

[0023] Further, the process of path planning for the key points in the three-dimensional model, obtaining the laser cleaning path, and constructing the laser parameter control model to obtain the laser parameters of each key point in the laser cleaning path includes:

[0024] Using a global path planning algorithm to plan paths for the key points in the three-dimensional model to obtain the laser cleaning path, constructing a laser parameter control model based on deep learning, inputting the key features of each key point in the laser cleaning path and the key features of the background points adjacent to each key point into the laser parameter control model, and outputting the laser parameters of each key point in the laser cleaning path according to the laser parameter control model.

[0025] Further, the process of real-time monitoring and feedback correction of the laser cleaning operation on the workpiece to be cleaned and the laser parameter control model includes:

[0026] According to the laser cleaning path and the laser parameters of each key point in the laser cleaning path, the laser cleaning operation is performed on the workpiece to be cleaned. After the laser cleaning operation is completed, the texture feature set of each key point and the texture feature set of the background point adjacent to each key point are obtained. The texture feature set of each key point and the texture feature set of the background point adjacent to each key point are matched with the standard texture parameters to obtain the texture parameter similarity of each key point and each background point. The texture parameter similarity of each key point and each background point is compared with the preset texture parameter similarity threshold. If the texture parameter similarity of each key point and each background point is greater than or equal to the texture parameter similarity threshold, the laser cleaning is completed.

[0027] If the texture parameter similarity of the key point or the background point is less than the texture parameter similarity threshold, the key point or the background point is marked as an abnormal point. The key features and laser parameters of the abnormal point are fed back to the relevant operating personnel. The relevant operating personnel manually label the laser parameters of the abnormal point. According to the laser parameters of the manually labeled abnormal point, the workpiece to be cleaned is reactivated and cleaned. At the same time, the key features of the abnormal point and the manually labeled laser parameters are supplemented to the training set of the laser parameter control model. The laser parameter control model is retrained, and the laser parameter control model after retraining is output.

[0028] Compared with the prior art, the beneficial effects of the present application are:

[0029] 1. Improved cleaning efficiency and accuracy: By obtaining the geometric data of the workpiece through three-dimensional scanning and combining with the surface image data for comprehensive analysis, the area to be cleaned can be more accurately identified. The gradient, Gaussian curvature, average curvature and texture feature set obtained by point cloud analysis make the laser cleaning path planning more accurate. This not only reduces the invalid cleaning area, but also ensures the cleaning quality.

[0030] 2. Adaptive parameter setting is realized: Based on the key features (including gradient, Gaussian curvature, average curvature and texture features), the laser parameter control model is constructed, which can automatically adjust the laser parameters such as power density, pulse width, spot size, etc. according to different surface characteristics. This adaptability helps to avoid overheating damage or incomplete cleaning, thereby improving the cleaning quality and safety.

[0031] 3. Enhanced complex structure processing capability: For workpieces with complex geometry or irregular surfaces, traditional laser cleaning methods often fail to achieve the desired results. The present application effectively solves this problem by converting the three-dimensional model into a topological undirected graph and optimizing the path planning of key points, so that even surfaces with rich details or uneven surfaces can be efficiently cleaned.

[0032] 4. Improved real-time monitoring and feedback correction capability: During the cleaning process, the system can monitor the cleaning status of each key point in real time and dynamically adjust the laser parameters according to the actual cleaning effect. If some areas do not meet the expected standards, they will be marked as abnormal points and the operator will be notified to intervene, and the training set will be updated to optimize the model performance. This closed-loop control system ensures that the entire cleaning process is always in the best state.

[0033] 5. Reduced need for manual intervention: Through high degree of automation of visual guidance technology and intelligent algorithms, the influence of human factors on cleaning results is greatly reduced. Operators only need to make a small amount of manual adjustment when necessary, thereby saving a lot of time and labor costs. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 A schematic diagram of a semiconductor laser cleaning method based on visual guidance according to an embodiment of the present application. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the present application will be described in detail below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0036] As shown in Figure 1 A semiconductor laser cleaning method based on visual guidance includes the following steps:

[0037] Step s1: Collect three-dimensional geometric data and surface image data of the workpiece to be cleaned by a three-dimensional scanning terminal, and construct a three-dimensional model of the workpiece to be cleaned according to the three-dimensional geometric data;

[0038] Step s2: Perform point cloud analysis on the three-dimensional model to obtain the gradient, Gaussian curvature and average curvature of each point in the three-dimensional model, perform texture analysis on the surface image data of the workpiece to be cleaned to obtain the texture feature set of each point in the three-dimensional model, and mark the gradient, Gaussian curvature, average curvature and texture feature set of each point in the three-dimensional model as key features;

[0039] Step s3: converting the three-dimensional model into a topological undirected graph, marking key points and background points in the topological undirected graph, setting the weight mark of each undirected edge in the topological undirected graph, and dividing the nodes in the topological undirected graph into key points and background points;

[0040] Step s4: path planning for the key points in the three-dimensional model, obtaining a laser cleaning path, constructing a laser parameter control model, and obtaining laser parameters for each key point in the laser cleaning path;

[0041] Step s5: real-time monitoring and feedback correction of the laser cleaning operation of the workpiece to be cleaned and the laser parameter control model.

[0042] It should be further explained that, in the specific implementation process, the process of point cloud analysis on the three-dimensional model to obtain the gradient, Gaussian curvature and average curvature of each point in the three-dimensional model includes:

[0043] Obtain the point cloud data contained in the three-dimensional model, pre-process the point cloud data, the pre-processing includes point cloud cleaning, removing noise points or outliers, ensuring the data quality of subsequent processing, using uniform downsampling method to reduce the number of points in the point cloud to reduce the computational complexity, while maintaining the main geometric characteristics of the model, obtaining the normal vector of each point in the pre-processed point cloud data, the present application uses the NormalEstimation class in the PCL library to estimate the normal, defines a neighborhood for each point, the points in the neighborhood will be used for subsequent gradient calculation, the size of the neighborhood depends on the point cloud density and the required level of detail, the specific steps are not repeated, a local plane is fitted in the neighborhood of each point, the specific process of fitting a local plane in the neighborhood of each point is as follows: calculate the average position of all points in the neighborhood, i.e. the centroid C, take the centroid C as the origin, calculate the position vector of each point p i relative to the centroid C, then use these vectors to construct a 3x3 covariance matrix Σ, N represents the total number of points in the neighborhood, the covariance matrix Σ is eigenvalue decomposed to obtain three eigenvalues λ1, λ2, λ3 and corresponding unit eigenvectors v1, v2, v3, if v1 is the eigenvector corresponding to the smallest eigenvalue, then the eigenvector corresponding to the smallest eigenvalue v1 is the normal direction of the plane to be solved, using the normal vector v3 and the centroid C can represent the equation of the local plane ax+by+cz=d, wherein a, b, c are the components of the eigenvector corresponding to the smallest eigenvalue, and d is a constant term calculated by substituting the coordinates of any point on the plane, based on the local plane fitted by each point, the gradient of each point is obtained, and the Gaussian curvature and the average curvature of each point are obtained, wherein the Gaussian curvature and the average curvature of each point are obtained using the eigenvalues of the covariance matrix, Gaussian curvature=λ1×λ2, wherein λ1, λ2 are the two smallest eigenvalues of the covariance matrix Σ, and average curvature=(λ1+λ2) / 2.

[0044] It should be further explained that, in the specific implementation process, the process of performing texture analysis on the surface image data of the workpiece to be cleaned to obtain the texture feature set of each point in the three-dimensional model includes:

[0045] The surface image data of the workpiece to be cleaned is preprocessed. It should be noted that the surface image data of the workpiece to be cleaned may have noise problems. Therefore, in an embodiment of the present application, the surface image data is preprocessed after being obtained to improve the quality of the surface image data and obtain surface image data for subsequent analysis. The image preprocessing operation improves the image quality, which is a well-known image processing method to those skilled in the art. In the embodiment of the application, the noise in the image is removed by a Gaussian filtering method. Subsequently, the surface image data is equalized to improve the contrast within the image, and the surface image data is grayed to convert the color image to a grayscale image. The graying process is usually achieved by averaging the weights of the color channels. The noise equalization processing, semantic segmentation network and graying are well-known technical means to those skilled in the art, and the specific steps will not be described again. The gray level of the surface image data is quantized to discrete gray levels, the gray value range is divided into several levels, for example, an 8-bit image can be divided into 16, 32, 64 levels, and the parameters required for the gray level co-occurrence matrix are defined, including distance (d) and direction (θ), which are determined according to actual application requirements. The specific steps will not be described again. According to the gray level co-occurrence matrix of each pixel point, the texture feature set of each pixel point is obtained. Each pixel point in the surface image data is matched with each point in the point cloud data of the surface of the three-dimensional model to obtain the corresponding relationship between each pixel point and each point in the point cloud data. According to the corresponding relationship between each pixel point and each point in the point cloud data, the texture feature set of each point in the point cloud data is obtained. The texture features in the texture feature set include but are not limited to contrast, correlation, energy, homogeneity, entropy, etc.

[0046] It should be further explained that, in the specific implementation process, the process of converting the three-dimensional model into a topological undirected graph includes:

[0047] The key features corresponding to each point in the point cloud data contained in the three-dimensional model are obtained. The three-dimensional model is converted into a topological undirected graph. Each point in the point cloud data contained in the three-dimensional model is taken as a node of the topological undirected graph. The key features corresponding to each point in the point cloud data contained in the three-dimensional model are imported into the node corresponding to each point.

[0048] Key points and background points are marked in the topological undirected graph. Undirected edges are constructed between adjacent nodes in the topological undirected graph. Undirected edges are also constructed between each node and the key points, and between each node and the background points in the topological undirected graph.

[0049] It should be further explained that, in the specific implementation process, the process of setting the weight labels of the undirected edges in the topological undirected graph includes:

[0050] The key features between adjacent nodes are compared, the key feature similarity SIM between adjacent nodes is obtained, and the weight label D of the undirected edge between adjacent nodes is set according to the key feature similarity SIM between adjacent nodes, wherein, The key features of each node are compared with the key features of the key points and the background points, the key feature similarity between each node and the key points and the key feature similarity between each node and the background points are obtained, the weight label of the undirected edge between each node and the key points is set according to the key feature similarity between each node and the key points, and the weight label of the undirected edge between each node and the background points is set according to the key feature similarity between each node and the background points;

[0051] An energy function is constructed according to the weight labels of the undirected edges in the topological undirected graph, and the topological undirected graph is segmented based on the energy function by using a graph cut algorithm, the minimum cut algorithm in the graph cut algorithm is used for segmentation, the path from the source point to the sink point is cut off constantly until the sum of the weights of all edges that are cut off is minimum (that is, the energy function is minimized), then the node connected to the source point at this time is marked as a key point, and the node connected to the sink point is marked as a key point, and the nodes in the topological undirected graph are divided into key points and background points according to the segmentation result.

[0052] It should be further explained that the core of the graph cut algorithm is to find a minimum cut, that is, a segmentation method, so that the path from the source point to the sink point is cut off, and the cost (that is, the sum of the weights of all edges that are cut off) of this cut is minimum, once the energy function is defined, it can be converted into a minimum cut problem on the graph, then, by solving the minimum cut problem, a cut that divides the graph into two parts can be found, and this cut corresponds to the optimal segmentation of the image, wherein the calculation formula of the energy function is:

[0053] E(A)=∑ p∈P D (p，a) +∑ p∈P D (p，b) +γ∑ (p，q)∈N D (p，q) ;

[0054] Wherein, A represents a segmentation of the topological undirected graph, P is the set of all points, N is the set of all adjacent points, D (p，a) represents the weight label of the undirected edge between the point p and the key point a, D (p，b) represents the weight label of the undirected edge between the point p and the background point b, D (p，a) and D (p，b)A cost for assigning a node to a key point or a background point, which is calculated based on gradient, Gaussian curvature, average curvature and texture feature set, for example, if the average curvature of a point is very similar to the average curvature of a known key point, then the cost of marking it as a key point is small, ensuring that each point is correctly assigned to the corresponding category, D (p，q) A weight label representing an undirected edge between point p and point q, D (p，q) For encouraging the consistency of properties such as gradient, texture or curvature within the same region, if two adjacent nodes belong to the same region, it is expected that there is a small difference between them, thereby reducing the value of the energy function, ensuring the spatial consistency of the segmentation result, encouraging adjacent pixels or feature points to have similar labels to avoid producing unnatural segmentation boundaries, thereby helping to maintain the continuity and smoothness of the region.

[0055] It should be further explained that in the specific implementation process, the process of marking key points and background points in the topological undirected graph includes:

[0056] Obtain the standard texture parameters of the workpiece to be cleaned, perform feature matching between the texture feature set of each point in the topological undirected graph and the standard texture parameters, obtain the texture parameter similarity of each point with the standard texture parameters, screen out the point corresponding to the minimum texture parameter similarity, mark the point as a key point, and screen out the point corresponding to the maximum texture parameter similarity, mark the point as a background point.

[0057] It should be further explained that in the specific implementation process, the process of path planning for the key points in the three-dimensional model, obtaining the laser cleaning path, constructing the laser parameter control model, and obtaining the laser parameters of each key point in the laser cleaning path includes:

[0058] The global path planning algorithm (such as A* algorithm, RRT-Connect, etc.) is used to plan a path for the key points in the three-dimensional model, to obtain a laser cleaning path, to construct a laser parameter control model based on deep learning, to establish a virtual working environment on a computer based on the idea of imitation learning, to simulate the entire cleaning process, to obtain the corresponding best simulated laser parameters under the condition that different simulated key features of the key points and different simulated key features of the background points adjacent to the key points in the laser cleaning path, to take the best simulated laser parameters corresponding to the different simulated key features of the key points and the different simulated key features of the background points adjacent to the key points as the training set and the test set, to input the training set into the laser parameter control model for training until the loss function training is stable, to save the model parameters, to test the laser parameter control model through the test set until the preset requirements are met, to output the laser parameter control model, and to input the key features of each key point in the laser cleaning path and the key features of the background points adjacent to each key point into the laser parameter control model, and to output the laser parameters of each key point in the laser cleaning path according to the laser parameter control model.

[0059] It needs to be further explained that, in the specific implementation process, the process of real-time monitoring and feedback correction of the laser cleaning operation on the workpiece to be cleaned and the laser parameter control model includes:

[0060] According to the laser cleaning path and the laser parameters of each key point in the laser cleaning path, the laser cleaning operation is performed on the workpiece to be cleaned. After the laser cleaning operation is completed, the texture feature set of each key point and the texture feature set of the background points adjacent to each key point in the laser cleaning path are obtained, the texture feature set of each key point and the texture feature set of the background points adjacent to each key point are matched with the standard texture parameters, the texture parameter similarity of each key point and each background point is obtained, and the texture parameter similarity of each key point and each background point is compared with the preset texture parameter similarity threshold. If the texture parameter similarity of each key point and each background point is greater than or equal to the texture parameter similarity threshold, the laser cleaning is completed.

[0061] If the texture parameter similarity of the key point or the background point is less than the texture parameter similarity threshold, the key point or the background point is marked as an abnormal point, the key features and the laser parameters of the abnormal point are fed back to the relevant operating personnel, the laser parameter of the abnormal point is manually labeled by the relevant operating personnel, the workpiece to be cleaned is re-cleaned according to the laser parameter of the manually labeled abnormal point, and the key features of the abnormal point and the manually labeled laser parameters are supplemented to the training set of the laser parameter control model, the laser parameter control model is retrained, and the laser parameter control model after retraining is output.

[0062] The above examples are only used to illustrate the technical method of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present application.

Claims

1. A method of visual guidance-based semiconductor laser cleaning, characterized by, The method comprises the following steps: Step s1: acquiring three-dimensional geometric data and surface image data of the workpiece to be cleaned by a three-dimensional scanning terminal, and constructing a three-dimensional model of the workpiece to be cleaned according to the three-dimensional geometric data; Step s2: performing point cloud analysis on the three-dimensional model to obtain the gradient, Gaussian curvature and average curvature of each point in the three-dimensional model, and performing texture analysis on the surface image data of the workpiece to be cleaned to obtain a texture feature set of each point in the three-dimensional model, and marking the gradient, Gaussian curvature, average curvature and texture feature set of each point in the three-dimensional model as key features; Step s3: converting the three-dimensional model into a topological undirected graph, marking key points and background points in the topological undirected graph, setting weight labels of each undirected edge in the topological undirected graph, and dividing the nodes in the topological undirected graph into key points and background points; Step s4: planning a path for the key points in the three-dimensional model, obtaining a laser cleaning path, constructing a laser parameter control model, and obtaining laser parameters of each key point in the laser cleaning path; Step s5: real-time monitoring and feedback correction of the laser cleaning operation of the workpiece to be cleaned and the laser parameter control model; The process of setting weight labels of each undirected edge in the topological undirected graph and dividing the nodes in the topological undirected graph into key points and background points comprises: comparing the key features between adjacent nodes to obtain the similarity of the key features between adjacent nodes, setting the weight labels of the undirected edges between adjacent nodes according to the similarity of the key features between adjacent nodes, and comparing the key features of each node with the key features of key points and background points respectively to obtain the similarity of the key features between each node and key points and the similarity of the key features between each node and background points, setting the weight labels of the undirected edges between each node and key points according to the similarity of the key features between each node and key points, and setting the weight labels of the undirected edges between each node and background points according to the similarity of the key features between each node and background points; constructing an energy function according to the weight labels of each undirected edge in the topological undirected graph, segmenting the topological undirected graph based on the energy function by using a graph cut algorithm, and dividing the nodes in the topological undirected graph into key points and background points according to the segmentation result; The process of marking key points and background points in the topological undirected graph comprises: obtaining a standard texture parameter of the workpiece to be cleaned, performing feature matching between the texture feature set of each point in the topological undirected graph and the standard texture parameter to obtain a texture parameter similarity between each point and the standard texture parameter, screening out a point corresponding to the minimum texture parameter similarity, marking the point as a key point, and screening out a point corresponding to the maximum texture parameter similarity, marking the point as a background point.

2. The method of claim 1, wherein the method is a visual guidance based semiconductor laser cleaning method. The process of performing point cloud analysis on the three-dimensional model to obtain the gradient, Gaussian curvature and average curvature of each point in the three-dimensional model comprises: obtaining point cloud data contained in the three-dimensional model, preprocessing the point cloud data, obtaining a normal vector of each point in the preprocessed point cloud data, defining a neighborhood for each point, fitting a local plane in the neighborhood of each point, and obtaining the gradient, Gaussian curvature and average curvature of each point based on the fitted local plane of each point.

3. The method of claim 2, wherein the method is a visual guidance based semiconductor laser cleaning method. The process of performing texture analysis on the surface image data of the workpiece to be cleaned to obtain the texture feature set of each point in the three-dimensional model includes: The process of performing texture analysis on the surface image data of the workpiece to be cleaned to obtain the texture feature set of each point in the three-dimensional model includes:

4. The method of claim 3, wherein the method is a visual guidance based semiconductor laser cleaning method. The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes:

5. The method of claim 4, wherein the method is a visual guidance based semiconductor laser cleaning method. The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes:

6. The method of claim 5, wherein the method is a visual guidance based semiconductor laser cleaning method. The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model into a topological undirected graph includes: The process of converting the three-dimensional model If the texture parameter similarity of the key point or the background point is less than the texture parameter similarity threshold, the key point or the background point is marked as an abnormal point, the key feature and the laser parameter of the abnormal point are fed back to the relevant operator, the operator manually labels the laser parameter of the abnormal point, reactivates the cleaning operation on the workpiece to be cleaned according to the laser parameter of the manually labeled abnormal point, and supplements the key feature of the abnormal point and the manually labeled laser parameter to the training set of the laser parameter control model, re-trains the laser parameter control model, and outputs the laser parameter control model after re-training.

Images