Gaussian Modeling Method Combined with Laser Scanning Point Cloud Data Processing

Abstract

The present application provides a Gaussian modeling method combined with laser scanning point cloud data processing, which relates to the technical field of three-dimensional modeling, and includes: registering multiple point cloud data sets by using a point cloud fusion algorithm to obtain a target point cloud data set; respectively analyzing the data accuracy, position deviation, and noise influence of multiple point cloud data sets to determine the point cloud data accuracy, position deviation coefficient, and noise influence coefficient, and comprehensively evaluating to determine the point cloud acquisition accuracy; setting a data dimensionality reduction ratio according to the point cloud acquisition accuracy, and performing covariance matrix dimensionality reduction on the target point cloud data set; selecting a predetermined kernel function, performing data fitting and Gaussian process regression modeling on the dimensionality-reduced point cloud data set, and outputting a surface fitting model. Through the present application, the technical problem that the traditional Gaussian modeling method needs to calculate a large number of covariance matrices, resulting in excessive storage and computing resource requirements and unable to achieve efficient three-dimensional modeling can be solved; the dimensionality of the data can be effectively reduced, and the efficiency of three-dimensional modeling can be significantly improved.

Description

Technical Field

[0001] This application relates to the field of 3D modeling technology, and in particular to a Gaussian modeling method combined with laser scanning point cloud data processing. Background Art

[0002] With the wide application of laser scanning technology, point cloud data has become an important data source in the fields of 3D modeling and spatial analysis. In the fields of computer vision, remote sensing, building information modeling, geographic information system, etc., point cloud data is used to create high-precision 3D models.

[0003] Traditional Gaussian modeling methods based on point cloud data (such as Gaussian process regression) have been proven to be able to effectively perform smooth fitting and surface modeling on point cloud data. However, with the continuous increase in the scale of point cloud data, these traditional methods face huge computational and storage pressures when dealing with high-resolution, large-scale point cloud data. Summary of the Invention

[0004] The purpose of this application is to provide a Gaussian modeling method combined with laser scanning point cloud data processing to solve the technical problem that traditional Gaussian modeling methods based on point cloud data need to calculate a large number of covariance matrices, resulting in excessive storage and computational resource requirements and unable to achieve efficient 3D modeling.

[0005] In view of the above problems, this application provides a Gaussian modeling method combined with laser scanning point cloud data processing, including: performing multiple laser scans on the target object to obtain multiple point cloud data sets, using a multi-resolution point cloud fusion algorithm to perform data registration on the multiple point cloud data sets to obtain a target point cloud data set, where each point cloud data set corresponds to a scanning angle; respectively analyzing the data accuracy, position deviation, and noise influence of the multiple point cloud data sets to determine the point cloud data accuracy, position deviation coefficient, and noise influence coefficient, and comprehensively evaluating to determine the point cloud acquisition accuracy; setting a data dimensionality reduction ratio according to the point cloud acquisition accuracy, performing covariance matrix dimensionality reduction on the target point cloud data set to obtain a dimensionality-reduced point cloud data set; selecting a predetermined kernel function, performing data fitting and Gaussian process regression modeling on the dimensionality-reduced point cloud data set, and outputting a surface fitting model of the target object.

[0006] Optionally, the Gaussian modeling method combined with laser scanning point cloud data processing further includes: randomly selecting a first point cloud dataset at a first scanning angle from the multiple point cloud datasets, and calculating a first average point spacing and a first area coverage rate of the first point cloud dataset, and determining a first data accuracy according to the first average point spacing and the first area coverage rate; obtaining a first standard point cloud dataset of the target object at the first scanning angle, performing a similarity traversal comparison between the first standard point cloud dataset and the first point cloud dataset, and determining a first position deviation coefficient according to the comparison result; calculating a first point cloud standard deviation and a first signal-to-noise ratio of the first point cloud dataset, and determining a first noise influence coefficient according to the first point cloud standard deviation and the first signal-to-noise ratio; continuing to perform data accuracy, position deviation, and noise influence analysis on other point cloud datasets in the multiple point cloud datasets to obtain multiple point cloud data accuracies, multiple position deviation coefficients, and multiple noise influence coefficients.

[0007] Optionally, the Gaussian modeling method combined with laser scanning point cloud data processing further includes: randomly selecting any area in the first point cloud dataset as a first point cloud area according to a predetermined size feature, where the predetermined size feature is a 5-by-5 point cloud area; randomly selecting a first standard point cloud area in the first standard point cloud dataset according to the predetermined size feature; performing distribution feature extraction and distribution similarity analysis on the first point cloud area and the first standard point cloud area to obtain a first area similarity; continuing to randomly select and perform similarity analysis in the first standard point cloud dataset according to the predetermined size feature until a predetermined selection number is reached, outputting multiple standard point cloud areas and multiple area similarities, and selecting the standard point cloud area with the largest area similarity as a first mapped point cloud area; calculating a coordinate deviation distance between the first point cloud area and the first mapped point cloud area in the target object to obtain a first position deviation distance; continuing to calculate the position deviation distances of other point cloud areas in the first point cloud dataset to obtain multiple position deviation distances, and calculating the average value of the multiple position deviation distances to obtain the first position deviation coefficient.

[0008] Optionally, the Gaussian modeling method combined with laser scanning point cloud data processing further includes: performing dimensionless processing on the multiple point cloud data accuracies, multiple position deviation coefficients, and multiple noise influence coefficients to obtain multiple normalized point cloud data accuracies, multiple normalized position deviation coefficients, and multiple normalized noise influence coefficients; configuring parameter influence weights according to the influence degrees of the point cloud data accuracy, position deviation coefficient, and noise influence coefficient on the point cloud acquisition accuracy; configuring angle influence weights according to multiple scanning angles, and performing weighted calculation on the multiple normalized point cloud data accuracies, multiple normalized position deviation coefficients, and multiple normalized noise influence coefficients according to the angle influence weights and the parameter influence weights to output the point cloud acquisition accuracy.

[0009] Optionally, the Gaussian modeling method combined with laser scanning point cloud data processing further includes: pre-training a dimensionality reduction ratio analysis channel, inputting the point cloud acquisition accuracy into the dimensionality reduction ratio analysis channel, and outputting a data dimensionality reduction ratio; according to the data dimensionality reduction ratio, performing covariance matrix dimensionality reduction on the target point cloud data set, and outputting the dimensionality-reduced point cloud data set.

[0010] Optionally, the Gaussian modeling method combined with laser scanning point cloud data processing further includes: obtaining a sample point cloud acquisition accuracy set according to the point cloud acquisition records in the historical time zone, screening the data dimensionality reduction ratios under different sample point cloud acquisition accuracies, obtaining multiple data dimensionality reduction ratios that meet the predetermined modeling accuracy, calculating the average value to obtain the sample data dimensionality reduction ratio, and constructing a sample data dimensionality reduction ratio set; using the sample point cloud acquisition accuracy set and the sample data dimensionality reduction ratio set to perform supervised learning on the random forest until convergence, and obtaining a trained dimensionality reduction ratio analysis channel.

[0011] Optionally, the Gaussian modeling method combined with laser scanning point cloud data processing further includes: converting the target point cloud data set into a multi-dimensional point cloud matrix, calculating to obtain a target covariance matrix; determining the number of principal components according to the data dimensionality reduction ratio, and performing dimensionality reduction processing on the target covariance matrix by using the principal component analysis algorithm, and outputting the dimensionality-reduced point cloud data set.

[0012] Optionally, the Gaussian modeling method combined with laser scanning point cloud data processing further includes: selecting an adaptive kernel function according to the multiple normalized noise influence coefficients, and using the adaptive kernel function to calculate the similarity of the dimensionality-reduced point cloud data set to generate a dimensionality-reduced covariance matrix; performing data fitting and Gaussian process regression modeling on the dimensionality-reduced covariance matrix to generate a surface fitting model of the target object.

[0013] The technical solutions provided in this application have at least the following technical effects or advantages:

[0014] By performing multiple laser scans on the target object, multiple point cloud data sets are obtained. The multi-resolution point cloud fusion algorithm is used to perform data registration on the multiple point cloud data sets to obtain the target point cloud data set, where each point cloud data set corresponds to a scanning angle. Then, the data accuracy, position deviation, and noise impact of the multiple point cloud data sets are analyzed respectively to determine the point cloud data accuracy, position deviation coefficient, and noise impact coefficient, and the point cloud acquisition accuracy is comprehensively evaluated. Then, according to the point cloud acquisition accuracy, the data dimensionality reduction ratio is set, and the covariance matrix of the target point cloud data set is reduced in dimension to obtain the reduced-dimensional point cloud data set. Finally, a predetermined kernel function is selected to perform data fitting and Gaussian process regression modeling on the reduced-dimensional point cloud data set, and the surface fitting model of the target object is output. That is to say, by setting an appropriate data dimensionality reduction ratio according to the point cloud data acquisition accuracy, the dimension of the data can be effectively reduced while retaining the main features and information of the data, and the efficiency of 3D modeling can be significantly improved, so as to be able to process and model the target point cloud data efficiently and accurately, and output a continuous, smooth and high-precision target surface model.

[0015] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are specifically given. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easy to understand through the following description of the specification. Brief Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0017] Figure 1 It is a flowchart of a Gaussian modeling method combining laser scan point cloud data processing according to the present application;

[0018] Figure 2 It is a flowchart of determining the first position deviation coefficient in a Gaussian modeling method combining laser scan point cloud data processing according to the present application. Detailed Description of the Embodiments

[0019] By providing a Gaussian modeling method combined with laser scanning point cloud data processing, this application solves the technical problem that the traditional Gaussian modeling method based on point cloud data requires a large amount of covariance matrix calculations, resulting in excessively high storage and computational resource requirements and unable to achieve efficient three-dimensional modeling. By setting an appropriate data dimensionality reduction ratio according to the point cloud data acquisition accuracy, it is possible to effectively reduce the data dimensionality while retaining the main features and information of the data, significantly improving the efficiency of three-dimensional modeling, and thus being able to process and model the target point cloud data efficiently and accurately, outputting a continuous, smooth, and high-precision target surface model.

[0020] Next, the technical solutions in this application will be described clearly and completely with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the example embodiments described here. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application. Additionally, it should be noted that for the sake of description, only parts related to this application are shown in the drawings rather than all.

[0021] Embodiment, please refer to the attached Figure 1 , this application provides a Gaussian modeling method combined with laser scanning point cloud data processing, which specifically includes the following steps:

[0022] S100: Perform multiple laser scans on the target object to obtain multiple point cloud data sets, and use a multi-resolution point cloud fusion algorithm to register the multiple point cloud data sets to obtain a target point cloud data set, where each point cloud data set corresponds to a scanning angle.

[0023] Specifically, first, select a laser scanning device suitable for the target object, such as a lidar device or a 3D laser scanner, which can accurately obtain the three-dimensional point cloud data of the target object at different angles and distances; then configure the scanning device to ensure that the scanning accuracy, scanning range, and sampling rate of the laser scanning device meet the task requirements. The scanning device needs to be able to perform multiple scans in all directions of the target object to ensure the comprehensiveness and meticulousness of the data; then use the selected laser scanning device to perform multiple laser scans on the target object (such as a target building, a power grid tower, etc.), where the scanning angle is different for each laser scan, that is, according to the geometric shape and characteristics of the target object, design a reasonable scanning path and scanning angle to ensure that different scanning angles can cover all surfaces of the target object and avoid occlusion or blind spots; through multiple scans, obtain point cloud data from multiple angles. Each scan will generate a point cloud data set, which contains the spatial coordinates and other additional information (such as intensity, color, etc.) obtained from the scanning position and angle, and obtain multiple point cloud data sets.

[0024] Next, the multi-resolution point cloud fusion algorithm is used to register the multiple point cloud data sets. The multi-resolution point cloud fusion algorithm registers the point cloud at multiple resolution levels. First, coarse registration is performed, and then refined registration is carried out by gradually increasing the resolution. First, downsample the point cloud data. Use methods such as voxel grid method to reduce the number of points in the point cloud and reduce the data complexity, which makes the registration process more efficient, especially when computing resources are limited. Then, perform coarse registration on the downsampled point cloud data. Usually use feature matching-based methods to find the initial transformation (translation, rotation) between different point clouds. The main purpose of this step is to estimate the approximate pose (position and orientation). Then gradually restore the resolution of the point cloud. As the resolution increases, the details of the data set are gradually retained, and the registration accuracy also increases. Further, use higher-resolution point cloud data for registration. Based on the initial pose obtained from the coarse registration, perform fine alignment on the point cloud. This can accurately calculate the relative transformation between each point cloud through iterative algorithms (such as ICP, NDT, etc.), eliminate errors, and obtain a more accurate registration result. The point cloud data obtained at multiple scanning angles can be fused into a unified point cloud data set after registration. During the fusion process, it may be necessary to remove duplicate points, fill in missing areas or perform interpolation processing to generate a complete target point cloud data set. Finally, after multiple laser scans and processing by the multi-resolution point cloud fusion algorithm, a unified target point cloud data set is obtained. This data set is generated by effectively registering and fusing multiple point cloud data sets, contains information from different scanning angles, and has high accuracy and integrity.

[0025] By combining the multi-resolution point cloud fusion algorithm, the problems of registration and fusion of point cloud data obtained by multiple laser scans are effectively solved, providing an efficient and accurate way to process point cloud data, and finally generating a unified and high-quality target point cloud data set.

[0026] S200: Analyze the data accuracy, position deviation and noise influence of the multiple point cloud data sets respectively, determine the point cloud data accuracy, position deviation coefficient and noise influence coefficient, and comprehensively evaluate to determine the point cloud acquisition accuracy.

[0027] Furthermore, step S200 of this application further includes:

[0028] S210: Randomly select the first point cloud data set at the first scanning angle from the multiple point cloud data sets, and calculate the first average point spacing and the first area coverage rate of the first point cloud data set. Determine the first data accuracy according to the first average point spacing and the first area coverage rate.

[0029] Specifically, first randomly select a dataset from multiple point cloud datasets, and this dataset corresponds to the first scanning angle of the target object (any one of multiple scanning angles); then calculate the first average point spacing and the first area coverage rate of the first point cloud dataset. The average point spacing refers to calculating the distance between every two points for all points in the first point cloud dataset and obtaining the average value of all distances. The first average point spacing reflects the density and accuracy of the point cloud data. A smaller average point spacing means a higher density and better accuracy of the point cloud data. The area coverage rate of the point cloud data is used to measure the proportion of the covered area in the scanned data and the uniformity of the points within the area. The higher the area coverage rate, the more complete the area covered by the data at this scanning angle and the higher the accuracy. The area coverage rate usually refers to the point distribution of the point cloud data within a certain area. For example, first project the point cloud dataset onto a two-dimensional plane and divide the area into grids; then count the number of points in each grid cell and calculate the filling degree of the grid cell (i.e., the density of points in each grid). The coverage rate is the ratio of the number of grid cells filled with points in the point cloud data to the total number of grid cells. Among them, the covered grid cells refer to the grid cells containing points, and the first area coverage rate is obtained. The first area coverage rate reflects the distribution range and uniformity of the point cloud dataset in space. A high coverage rate means that the data can comprehensively cover the surface of the target object with fewer missing areas.

[0030] Then, combine the first average point spacing and the first area coverage rate to comprehensively evaluate the data accuracy of the point cloud data. For example, a smaller average point spacing and a higher area coverage rate usually indicate higher data accuracy. Combine the first average point spacing and the first area coverage rate to define a comprehensive index (such as a weighted average) to calculate the data accuracy. If the point cloud data has a high density and a complete covered area, the data accuracy is high; if the cloud data is sparse or the covered area is incomplete, the data accuracy is low. By calculating the first average point spacing and the first area coverage rate for the first point cloud dataset, the accuracy of the dataset can be comprehensively evaluated, providing a basis for subsequent point cloud data processing, model construction, and optimization.

[0031] S220: Obtain the first standard point cloud dataset of the target object at the first scanning angle, perform a similarity traversal comparison between the first standard point cloud dataset and the first point cloud dataset, and determine the first position deviation coefficient according to the comparison result.

[0032] Furthermore, as Figure 2 shown, step S220 of this application further includes:

[0033] S221: Randomly select any area in the first point cloud dataset as the first point cloud area according to a predetermined size feature, where the predetermined size feature is a 5-by-5 point cloud area; S222: Randomly box-select in the first standard point cloud dataset according to the predetermined size feature to obtain the first standard point cloud area; S223: Extract distribution features and perform distribution similarity analysis on the first point cloud area and the first standard point cloud area to obtain the first area similarity; S224: Continue to randomly box-select and perform similarity analysis in the first standard point cloud dataset according to the predetermined size feature until a predetermined number of box-selections is reached, output multiple standard point cloud areas and multiple area similarities, and select the standard point cloud area with the maximum area similarity as the first mapped point cloud area; S225: Calculate the coordinate deviation distance between the first point cloud area and the first mapped point cloud area in the target object to obtain the first position deviation distance; S226: Continue to calculate the position deviation distances for other point cloud areas in the first point cloud dataset to obtain multiple position deviation distances, and calculate the average of the multiple position deviation distances to obtain the first position deviation coefficient.

[0034] Specifically, first, obtain a predetermined size feature, where the predetermined size feature is a 5-by-5 point cloud area, and the size of this area is determined by the spatial coordinates of each point in the point cloud data; then, according to the predetermined size feature, randomly select any area in the first point cloud dataset that meets the predetermined size feature as the first point cloud area. Next, randomly box-select in the first standard point cloud dataset according to the predetermined size feature to obtain the first standard point cloud area that meets the predetermined size feature. Then, extract the distribution features of the first point cloud area and the first standard point cloud area respectively. The distribution features may include the density of points, the local curvature of the point cloud, the spatial distribution pattern of the point cloud, etc.; and use the Pearson correlation coefficient to evaluate the similarity between the two areas to obtain the first area similarity.

[0035] Next, continue to randomly box-select and perform similarity analysis in the first standard point cloud dataset according to the predetermined size feature, that is, according to the predetermined size feature, continue to perform multiple random box-selections in the first standard point cloud dataset, and repeat the extraction of distribution features and similarity analysis for the standard point cloud area obtained by each box-selection until a predetermined number of box-selections is reached (which can be set according to the actual scenario, such as 1000 times), output multiple standard point cloud areas and multiple area similarities; then select the standard point cloud area with the maximum area similarity as the first mapped point cloud area. Further calculate the coordinate deviation distance between the first point cloud area and the first mapped point cloud area in three-dimensional space, and use the Euclidean distance to measure the difference between the two to obtain the first position deviation distance.

[0036] Further, for other point cloud regions in the first point cloud dataset, continue to calculate the deviation distances between them and the corresponding mapped regions according to the same method, and calculate the average value of multiple position deviation distances to obtain a first position deviation coefficient, that is, the average position deviation of the point cloud dataset.

[0037] S230: Calculate the first point cloud standard deviation and the first signal-to-noise ratio of the first point cloud dataset, and determine the first noise influence coefficient according to the first point cloud standard deviation and the first signal-to-noise ratio; S240: Continue to perform data accuracy, position deviation, and noise influence analysis on other point cloud datasets in the multiple point cloud datasets to obtain multiple point cloud data accuracies, multiple position deviation coefficients, and multiple noise influence coefficients.

[0038] Specifically, each point in the point cloud dataset has a spatial coordinate and may have errors or noise. To evaluate these errors, the standard deviation of the point cloud dataset can be calculated; then calculate the first point cloud standard deviation of the first point cloud dataset. The comprehensive standard deviation index of the three-dimensional point cloud can be used to calculate the overall standard deviation. The first point cloud standard deviation represents the overall dispersion or noise level of the point cloud dataset. The signal-to-noise ratio (SNR) is used to measure the ratio of the signal to the noise. The stronger the signal, the smaller the influence relative to the noise. For point cloud data, the signal can be the true position of the point cloud, and the noise is the error caused by the measurement error. Calculate the first signal-to-noise ratio of the first point cloud dataset. Finally, determine the first noise influence coefficient according to the first point cloud standard deviation and the first signal-to-noise ratio. The noise influence coefficient reflects the degree of influence of the noise in the point cloud data on the data quality. The larger the noise influence coefficient, the smaller the influence of the noise in the data, and the better the data quality.

[0039] Then, use the same method as above to obtain the first data accuracy, the first position deviation coefficient, and the first noise influence coefficient, and continue to perform data accuracy, position deviation, and noise influence analysis on other point cloud datasets in the multiple point cloud datasets to obtain multiple point cloud data accuracies, multiple position deviation coefficients, and multiple noise influence coefficients of the multiple point cloud datasets.

[0040] Furthermore, step S200 of the present application further includes:

[0041] S250: Perform dimensionless processing on the accuracies of the multiple point cloud data, the multiple position deviation coefficients, and the multiple noise influence coefficients to obtain multiple normalized point cloud data accuracies, multiple normalized position deviation coefficients, and multiple normalized noise influence coefficients; S260: Configure parameter influence weights according to the influence degrees of the point cloud data accuracy, the position deviation coefficient, and the noise influence coefficient on the point cloud acquisition accuracy; S270: Configure angle influence weights according to multiple scanning angles, and perform weighted calculation on the multiple normalized point cloud data accuracies, the multiple normalized position deviation coefficients, and the multiple normalized noise influence coefficients according to the angle influence weights and the parameter influence weights, and output the point cloud acquisition accuracy.

[0042] Specifically, first, perform dimensionless processing on the accuracies of the multiple point cloud data, the multiple position deviation coefficients, and the multiple noise influence coefficients. The purpose is to eliminate the influence between different dimensions and units, so that each index has a unified standard, which is convenient for subsequent comprehensive analysis and weighted calculation. For example, normalize the point cloud data accuracy through standard deviation normalization to make its range fall between 0 and 1; obtain multiple normalized point cloud data accuracies, multiple normalized position deviation coefficients, and multiple normalized noise influence coefficients. Through dimensionless processing, ensure that the ranges of all data are on the same scale, so that subsequent weighted calculation and influence degree analysis can be carried out reasonably, and each coefficient is expressed on a unified scale, which is convenient for comprehensive evaluation.

[0043] Next, configure and weight the influence degrees of different indexes (data accuracy, position deviation, noise influence) on the acquisition accuracy. The influence degree of each index on the acquisition accuracy can be assigned according to its importance in actual applications to obtain parameter influence weights. For example, configure the weights of each index through expert experience. Common weight configuration methods include manual configuration based on experience and automatic optimization based on data analysis. By reasonably configuring weights, the contribution of each index to the point cloud acquisition accuracy can be quantified, and ensure that the comprehensive evaluation result accurately reflects the quality of the point cloud data.

[0044] On the other hand, since the point cloud data is derived from multiple scanning angles, the description ability of the point cloud data at each scanning angle for the target object may vary. Generally speaking, certain angles may have better capturing ability for certain details of the target object, while other angles may have occlusion or large errors. According to the influence of multiple scanning angles on the target object, configure the angle influence weight for each scanning angle. For example, if the data at a certain angle is more accurate or has a higher coverage, a higher weight can be configured for it. Through experience or data analysis (such as based on factors such as the area size and density of the point cloud coverage), assign a weight to each scanning angle. The angle weight can be adjusted based on factors such as the scanning position, viewing angle, data coverage rate, etc. Then, according to the angle influence weight and the parameter influence weight, perform weighted calculations on the multiple normalized point cloud data accuracies, multiple normalized position deviation coefficients, and multiple normalized noise influence coefficients, and output the point cloud acquisition accuracy. By comprehensively considering the influence weights of multiple scanning angles and the influence weights of various data accuracy parameters, the overall point cloud acquisition accuracy can be accurately evaluated, and finally a reliable point cloud acquisition accuracy result can be output, which is convenient for subsequent point cloud data fusion, modeling, and optimization.

[0045] S300: Set the data dimensionality reduction ratio according to the point cloud acquisition accuracy, perform covariance matrix dimensionality reduction on the target point cloud dataset, and obtain the dimensionality-reduced point cloud dataset.

[0046] Furthermore, step S300 of the present application further includes:

[0047] S310: Pre-train the dimensionality reduction ratio analysis channel, input the point cloud acquisition accuracy into the dimensionality reduction ratio analysis channel, and output the data dimensionality reduction ratio.

[0048] Furthermore, step S310 of the present application further includes:

[0049] S311: According to the point cloud acquisition records in the historical time zone, obtain the sample point cloud acquisition accuracy set, screen the data dimensionality reduction ratios under different sample point cloud acquisition accuracies, obtain multiple data dimensionality reduction ratios that meet the predetermined modeling accuracy, calculate the mean value to obtain the sample data dimensionality reduction ratio, and construct the sample data dimensionality reduction ratio set; S312: Use the sample point cloud acquisition accuracy set and the sample data dimensionality reduction ratio set to perform supervised learning on the random forest until convergence, and obtain the trained dimensionality reduction ratio analysis channel.

[0050] Specifically, first, according to the point cloud acquisition records in the historical time zone (such as in the most recent three months), collect the acquisition accuracies of multiple point cloud data sets. These data may come from different scanning tasks, and each task contains information such as acquisition accuracy, scanning angle, noise level, etc.; then, for each acquisition record, the acquisition accuracy can be quantified by calculating the standard deviation, position deviation, noise impact coefficient, etc. of the data, ensuring that the acquisition accuracy set contains the acquisition accuracy information of the point cloud data under different conditions; then, based on the point cloud acquisition accuracy, for each sample point cloud data set, perform dimensionality reduction processing, and according to the target modeling accuracy requirements, select an appropriate dimensionality reduction ratio, that is, select a data dimensionality reduction ratio suitable for its acquisition accuracy for each sample data set, ensuring that the data after dimensionality reduction can meet the accuracy requirements in the subsequent modeling process. Through the analysis of multiple sample point cloud data sets, a series of data dimensionality reduction ratios suitable for different acquisition accuracies are obtained, and the mean value of multiple data dimensionality reduction ratios is calculated to obtain the sample data dimensionality reduction ratio. Then, all the dimensionality reduction ratios that meet the predetermined modeling accuracy requirements are integrated into a set to form the sample data dimensionality reduction ratio set.

[0051] Then, use the sample point cloud acquisition accuracy set and the sample data dimensionality reduction ratio set as sample training data, and use the random forest regression or classification model to perform supervised learning on the sample data. The random forest models the data through multiple decision trees and can effectively learn the patterns from complex data sets. In this step, the input feature is the sample point cloud acquisition accuracy, and the target variable is the corresponding dimensionality reduction ratio; among them, the random forest model will learn the corresponding dimensionality reduction ratio under different acquisition accuracies according to the accuracy information of the sample point cloud data. During the training process, the model optimizes its prediction ability through multiple iterations until the model converges, that is, reaches the predetermined accuracy or stop condition; the random forest algorithm will gradually adjust the parameters based on the training data set until convergence. Usually, methods such as cross-validation are used to evaluate the generalization ability of the model and ensure that the training process can effectively fit the data. When the training error of the model meets the preset threshold and further training cannot significantly improve the performance, the model is considered to be convergent, and the dimensionality reduction ratio analysis channel after training is obtained. Through the supervised learning of the random forest, a dimensionality reduction ratio analysis channel is obtained, which can automatically determine the appropriate dimensionality reduction ratio according to the point cloud acquisition accuracy, thereby improving the efficiency and accuracy of point cloud data modeling.

[0052] Finally, input the point cloud acquisition accuracy into the dimensionality reduction ratio analysis channel for analysis, and output the data dimensionality reduction ratio.

[0053] S320: According to the data dimensionality reduction ratio, perform covariance matrix dimensionality reduction on the target point cloud data set, and output the dimensionality-reduced point cloud data set.

[0054] Furthermore, step S320 of the present application further includes:

[0055] S321: Convert the target point cloud data set into a multi-dimensional point cloud matrix, and calculate the target covariance matrix; S322: Determine the number of principal components according to the data dimensionality reduction ratio, and use the principal component analysis algorithm to perform dimensionality reduction processing on the target covariance matrix, and output the dimensionality-reduced point cloud data set.

[0056] Specifically, first, convert the target point cloud data set into a multi-dimensional point cloud matrix. For example, assume that the target point cloud data set contains N points, and each point has D-dimensional features. In order to apply principal component analysis to reduce the dimensionality of the point cloud data, it is first necessary to convert the point cloud data set into a multi-dimensional matrix. For example, convert the target point cloud data set into an N×D matrix X, where each row represents a point and contains the D-dimensional coordinates of the point. The covariance matrix describes the correlation between each dimension in the point cloud data set. When calculating the covariance matrix, it is first necessary to centralize the data, that is, subtract the mean of each dimension from each dimension. The covariance matrix reveals the linear relationship between each dimension and helps the subsequent dimensionality reduction steps.

[0057] Next, according to the data dimensionality reduction ratio (for example, retaining 60% of the variance), determine the number of principal components k to be retained in the principal component analysis. The number of retained principal components determines the dimension after data dimensionality reduction. First, perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues and eigenvectors. The eigenvalues reflect the variance size of each principal component. Therefore, the number of retained principal components can be determined by the cumulative proportion of the variance occupied by the eigenvalues. Then, according to the number of principal components, use the principal component analysis algorithm to perform dimensionality reduction processing on the target covariance matrix. That is, after determining the number of principal components k, the first k eigenvectors can be selected to construct a dimensionality reduction matrix; project the original data matrix onto these principal components to obtain the dimensionality-reduced data set; the dimensionality-reduced data set retains the main features in the data while reducing the dimension. By performing dimensionality reduction on the point cloud data, the computational complexity is reduced, and sufficient information is retained to ensure efficient and accurate 3D modeling and data processing.

[0058] S400: Select a predetermined kernel function, perform data fitting and Gaussian process regression modeling on the dimensionality-reduced point cloud data set, and output the surface fitting model of the target object.

[0059] Furthermore, step S400 of the present application further includes:

[0060] S410: Select an adapted kernel function according to the multiple normalized noise influence coefficients, and use the adapted kernel function to calculate the similarity of the dimensionality-reduced point cloud data set to generate a dimensionality-reduced covariance matrix; S420: Perform data fitting and Gaussian process regression modeling on the dimensionality-reduced covariance matrix to generate the surface fitting model of the target object.

[0061] Specifically, first, an adaptive kernel function is selected according to the multiple normalized noise influence coefficients. For example, according to the normalized noise influence coefficients, a suitable kernel function is selected to adapt to the point cloud data set. For instance, for point cloud data with less noise, the RBF kernel may be selected, while for data with more noise, the Matern kernel or other kernel functions suitable for data with stronger noise may be selected. Then, using the adaptive kernel function, the similarity between each pair of points in the dimensionality-reduced point cloud data set is calculated, and based on the result of the similarity calculation, a covariance matrix after dimensionality reduction is constructed. The covariance matrix describes the similarity between each pair of points in the point cloud data set and is the basis for subsequent Gaussian process regression modeling.

[0062] Then, the dimensionality-reduced covariance matrix and the Gaussian process regression (GPR) model are used to fit the point cloud data. Gaussian process regression is a non-parametric method that predicts output values by calculating the similarity (covariance matrix) between input data points. Using the Gaussian process regression method to fit the point cloud data, a surface fitting model is generated. This model can not only provide predicted values of the point cloud surface but also give the prediction uncertainty (i.e., covariance) of each point, which is used to evaluate the reliability of the surface fitting. Through the Gaussian process regression model, a three-dimensional surface fitting model of the target object is generated. This model reflects the trend and shape of the point cloud data and can also provide a measure of uncertainty for each predicted point. Through Gaussian process regression modeling, a smooth and continuous surface fitting model of the target object is generated.

[0063] In summary, the Gaussian modeling method provided by this application in combination with laser scanning point cloud data processing has the following technical effects:

[0064] By performing multiple laser scans on the target object, multiple point cloud data sets are obtained. The multi-resolution point cloud fusion algorithm is used to register the multiple point cloud data sets to obtain the target point cloud data set, where each point cloud data set corresponds to a scanning angle. Then, the data accuracy, position deviation, and noise influence of the multiple point cloud data sets are analyzed respectively to determine the point cloud data accuracy, position deviation coefficient, and noise influence coefficient, and the point cloud acquisition accuracy is comprehensively evaluated. Then, according to the point cloud acquisition accuracy, the data dimensionality reduction ratio is set, and the covariance matrix of the target point cloud data set is reduced in dimension to obtain the dimensionality-reduced point cloud data set. Finally, a predetermined kernel function is selected to perform data fitting and Gaussian process regression modeling on the dimensionality-reduced point cloud data set, and the surface fitting model of the target object is output. That is to say, by setting an appropriate data dimensionality reduction ratio according to the point cloud data acquisition accuracy, the dimension of the data can be effectively reduced while retaining the main features and information of the data, significantly improving the efficiency of three-dimensional modeling, so as to be able to process and model the target point cloud data efficiently and accurately, and output a continuous, smooth, and high-precision target surface model.

[0065] The foregoing description of the disclosed embodiments enables those skilled in the art to practice or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Thus, the present application is not intended to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0066] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is also intended to include these changes and modifications.

Claims

1. Gaussian modeling method combined with laser scanning point cloud data processing, characterized in that, The method includes: Performing multiple laser scans on a target object to obtain multiple point cloud data sets, and registering the multiple point cloud data sets by using a multi-resolution point cloud fusion algorithm to obtain a target point cloud data set, where each point cloud data set corresponds to a scanning angle; Analyzing the data accuracy, position deviation, and noise influence of the multiple point cloud data sets respectively, determining the point cloud data accuracy, position deviation coefficient, and noise influence coefficient, and comprehensively evaluating to determine the point cloud acquisition accuracy; Setting a data dimensionality reduction ratio according to the point cloud acquisition accuracy, and performing covariance matrix dimensionality reduction on the target point cloud data set to obtain a dimensionality-reduced point cloud data set; Selecting a predetermined kernel function, performing data fitting and Gaussian process regression modeling on the dimensionality-reduced point cloud data set, and outputting a surface fitting model of the target object; Analyzing the data accuracy, position deviation, and noise influence of the multiple point cloud data sets respectively, including: Randomly selecting a first point cloud data set at a first scanning angle from the multiple point cloud data sets, and calculating the first average point spacing and the first area coverage rate of the first point cloud data set, and determining the first data accuracy according to the first average point spacing and the first area coverage rate; Obtaining a first standard point cloud data set of the target object at the first scanning angle, performing similar traversal comparison on the first standard point cloud data set and the first point cloud data set, and determining the first position deviation coefficient according to the comparison result; Calculating the first point cloud standard deviation and the first signal-to-noise ratio of the first point cloud data set, and determining the first noise influence coefficient according to the first point cloud standard deviation and the first signal-to-noise ratio; Continuing to analyze the data accuracy, position deviation, and noise influence of other point cloud data sets in the multiple point cloud data sets to obtain multiple point cloud data accuracies, multiple position deviation coefficients, and multiple noise influence coefficients.

2. The Gaussian modeling method for processing laser scanning point cloud data according to claim 1, characterized in that Performing similar traversal comparison on the first standard point cloud data set and the first point cloud data set, and determining the first position deviation coefficient according to the comparison result, including: Randomly selecting any area in the first point cloud data set as a first point cloud area according to a predetermined size feature, where the predetermined size feature is a 5-by-5 point cloud area; Performing random box selection in the first standard point cloud data set according to the predetermined size feature to obtain a first standard point cloud area; Performing distribution feature extraction and distribution similarity analysis on the first point cloud area and the first standard point cloud area to obtain a first area similarity; Continuing to perform random box selection and similarity analysis in the first standard point cloud data set according to the predetermined size feature until a predetermined number of box selection times is reached, outputting multiple standard point cloud areas and multiple area similarities, and selecting the standard point cloud area with the maximum area similarity as a first mapped point cloud area; Calculating the coordinate deviation distance between the first point cloud area and the first mapped point cloud area in the target object to obtain a first position deviation distance; Continuing to calculate the position deviation distances of other point cloud areas in the first point cloud data set to obtain multiple position deviation distances, and calculating the mean value of the multiple position deviation distances to obtain the first position deviation coefficient.

3. The Gaussian modeling method combined with laser scanning point cloud data processing according to claim 2, wherein Comprehensively evaluating to determine the point cloud acquisition accuracy, including: Perform dimensionless processing on the multiple point cloud data precisions, multiple position deviation coefficients, and multiple noise influence coefficients to obtain multiple normalized point cloud data precisions, multiple normalized position deviation coefficients, and multiple normalized noise influence coefficients; Configure parameter influence weights according to the influence degrees of the point cloud data precision, position deviation coefficient, and noise influence coefficient on the point cloud acquisition precision; Configure angle influence weights according to multiple scanning angles, and perform weighted calculation on the multiple normalized point cloud data precisions, multiple normalized position deviation coefficients, and multiple normalized noise influence coefficients according to the angle influence weights and parameter influence weights, and output the point cloud acquisition precision.

4. The Gaussian modeling method for processing combined laser scanning point cloud data according to claim 1, wherein, Set a data dimensionality reduction ratio according to the point cloud acquisition precision, and perform covariance matrix dimensionality reduction on the target point cloud data set to obtain a dimensionality-reduced point cloud data set, including: Pre-train a dimensionality reduction ratio analysis channel, and input the point cloud acquisition precision into the dimensionality reduction ratio analysis channel to output a data dimensionality reduction ratio; Perform covariance matrix dimensionality reduction on the target point cloud data set according to the data dimensionality reduction ratio, and output the dimensionality-reduced point cloud data set.

5. The Gaussian modeling method combined with laser scanning point cloud data processing according to claim 4, characterized in that, Pre-train a dimensionality reduction ratio analysis channel, including: Obtain a sample point cloud acquisition precision set according to the point cloud acquisition records in the historical time zone, and screen the data dimensionality reduction ratios under different sample point cloud acquisition precisions to obtain multiple data dimensionality reduction ratios that meet the predetermined modeling precision, calculate the mean value to obtain the sample data dimensionality reduction ratio, and construct a sample data dimensionality reduction ratio set; Use the sample point cloud acquisition precision set and the sample data dimensionality reduction ratio set to perform supervised learning on the random forest until convergence, and obtain a trained dimensionality reduction ratio analysis channel.

6. The Gaussian modeling method combined with laser scanning point cloud data processing according to claim 4, characterized in that Perform covariance matrix dimensionality reduction on the target point cloud data set according to the data dimensionality reduction ratio, and output the dimensionality-reduced point cloud data set, including: Convert the target point cloud data set into a multi-dimensional point cloud matrix, and calculate to obtain a target covariance matrix; Determine the number of principal components according to the data dimensionality reduction ratio, and use the principal component analysis algorithm to perform dimensionality reduction processing on the target covariance matrix, and output the dimensionality-reduced point cloud data set.

7. The Gaussian modeling method combined with laser scanning point cloud data processing according to claim 3, characterized in that Select a predetermined kernel function, perform data fitting and Gaussian process regression modeling on the dimensionality-reduced point cloud data set, and output a surface fitting model of the target object, including: Select an adapted kernel function according to the multiple normalized noise influence coefficients, and use the adapted kernel function to calculate the similarity of the dimensionality-reduced point cloud data set to generate a dimensionality-reduced covariance matrix; Perform data fitting and Gaussian process regression modeling on the dimensionality-reduced covariance matrix to generate the surface fitting model of the target object.

Images