Self-adaptive determination method and device for point cloud segmentation threshold parameter of stone-laying body

Abstract

The invention provides a stone-laying body point cloud segmentation threshold parameter adaptive determination method and device, and relates to the technical field of computers. Comprising the following steps: determining at least one sample area from point cloud data of a to-be-processed stone-laying body; according to the point cloud data of each sample area, curvature information and normal vector included angle parameters corresponding to the to-be-processed stone-laying body are determined; determining a first segmentation parameter according to the curvature information; determining a second segmentation parameter according to the normal vector included angle parameter; and taking the first segmentation parameter and the second segmentation parameter as point cloud segmentation parameters of the stone-laying body, and performing point cloud segmentation on the point cloud data of the stone-laying body to be processed according to the point cloud segmentation parameters. According to the method, the curvature information and the normal vector included angle information of the point cloud data can be extracted on the basis of the point cloud data characteristics of the stone masonry, so that the curvature threshold and the normal vector included angle threshold can be automatically and adaptively determined, and the precision and robustness of point cloud segmentation of the stone masonry are improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and more specifically, to a method and apparatus for adaptively determining the point cloud segmentation threshold parameters of masonry. Background Technology

[0002] In point cloud data processing, effective segmentation of stone units and masonry joints is a crucial prerequisite. Currently, point cloud segmentation methods based on region growing algorithms have become one of the mainstream technologies for processing point cloud data of regular stone masonry due to their clear principles and strong adaptability to regular geometric surfaces.

[0003] Currently, when segmenting point clouds of masonry structures using region growing algorithms, the segmentation threshold parameters typically rely on the prior knowledge of engineers and are implemented through manual trial and error. Operators preset an initial parameter based on their own experience and then repeatedly adjust and verify it by observing the segmentation results.

[0004] The above method results in low segmentation efficiency and poor accuracy of the point cloud of the masonry. Summary of the Invention

[0005] The purpose of this application is to address the shortcomings of the prior art by providing an adaptive method and apparatus for determining the point cloud segmentation threshold parameters of masonry, thereby achieving adaptive determination of the segmentation parameters of masonry and improving segmentation accuracy.

[0006] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide an adaptive method for determining the point cloud segmentation threshold parameter of masonry, including: At least one sample region is determined from the point cloud data of the masonry body to be processed, the sample region including the regular stone surface, the gaps between stones and the edges of the stones; Based on the point cloud data of each sample area, determine the curvature information and the angle parameter of the normal vector corresponding to the masonry body to be processed; Based on the curvature information, the first segmentation parameter is determined; The second segmentation parameter is determined based on the angle parameter between the normal vectors. The first segmentation parameter and the second segmentation parameter are used as point cloud segmentation parameters for the masonry body, and the point cloud data of the masonry body to be processed is segmented according to the point cloud segmentation parameters.

[0007] Optionally, determining the curvature information and normal vector angle parameters corresponding to the masonry body to be processed based on the point cloud data of each sample region includes: Based on the point cloud data of each sample region, determine the curvature information of each point in each sample region; The curvature information of each point in each sample region is used as the curvature information of the masonry body to be processed. Multiple sub-regions are determined from each sample region, and multiple pairs of points are determined for each sub-region; Based on multiple pairs of points corresponding to each sub-region in each sample region, determine the mean and standard deviation of the angle between the normal vectors; Candidate regions are selected from each sample region, and the set of normal vectors corresponding to each candidate region is determined. Based on the set of normal vectors corresponding to each candidate region in each sample region, determine the reference value of the angle between the normal vectors; The mean and standard deviation of the included angle of the normal vectors, as well as the reference value of the included angle of the normal vectors, are used as the included angle parameters of the normal vectors corresponding to the masonry body to be processed.

[0008] Optionally, determining the first segmentation parameter based on the curvature information includes: Based on the curvature information of each point in each sample region, determine the positions of the first curvature peak and the second curvature peak; The target valley location is determined based on the positions of the first and second curvature peaks. The first segmentation parameter is determined based on the curvature information of the target valley bottom location and a preset safety factor.

[0009] Optionally, determining the second segmentation parameter based on the angle parameter between the normal vectors includes: The first threshold is determined based on the mean and standard deviation of the angle between the normal vectors. Determine the second threshold based on the reference value of the angle between the normal vectors; Based on the first threshold and the second threshold, determine the value range of the second segmentation parameter; The second segmentation parameter is determined within the range of values ​​of the second segmentation parameter.

[0010] Optionally, determining multiple sub-regions from each sample region includes: Multiple initial sub-regions are determined from the sample region; Based on the number of points contained in each initial sub-region and the curvature information of each point, the initial sub-regions are first filtered to obtain each first-filtered sub-region. Principal component analysis is performed on each point in each first screening sub-region to determine the point set distribution characteristics of each first screening sub-region; Based on the point set distribution characteristics of each first screening sub-region, a second screening is performed on each first screening sub-region to obtain each second screening sub-region; Based on each of the second-selection sub-regions, a third selection is performed to obtain multiple sub-regions corresponding to the sample region.

[0011] Optionally, determining the mean and standard deviation of the angle between the normal vectors based on multiple pairs of points corresponding to each sub-region in each sample region includes: Based on multiple pairs of points corresponding to each sub-region in each sample region, the angle information of the normal vectors of each pair of points is determined, and the set of angles of the normal vectors corresponding to each sample region is obtained. Based on the information of the angles of the normal vectors in the set of angles of the normal vectors corresponding to each sample region, the mean and standard deviation of the angles of the normal vectors are determined.

[0012] Optionally, determining the set of normal vectors corresponding to each candidate region includes: Based on the centroid information of the candidate regions, cluster analysis is performed to determine the clusters corresponding to the candidate regions; Based on the distribution information and curvature information of each point in each cluster, the target sub-region in each cluster is determined; Based on the normal vector information of each point within the target sub-region in each cluster, the normal vector corresponding to each cluster is determined, and the normal vector corresponding to each cluster is used as the set of normal vectors corresponding to the candidate region.

[0013] Optionally, determining the reference value of the angle between the normal vectors based on the set of normal vectors corresponding to each candidate region in each sample region includes: Based on the normal vectors corresponding to each cluster in each candidate region of the sample region, the absolute value of the angle between the normal vectors of each pair of clusters is determined, and the set of angles between the normal vectors within the sample region is obtained. Based on the set of angles between the normal vectors within each sample region, a reference value for the angle between the normal vectors is determined.

[0014] Secondly, embodiments of this application also provide an adaptive determination device for point cloud segmentation threshold parameters of masonry, including: a determination module and a processing module; The determining module is used to determine at least one sample region from the point cloud data of the masonry body to be processed. The sample region includes the regular stone surface, the gaps between stones, and the edges of the stones. The determining module is used to determine the curvature information and normal vector angle parameters of the masonry body to be processed based on the point cloud data of each sample area. The determining module is used to determine the first segmentation parameter based on the curvature information; The determining module is used to determine the second segmentation parameter based on the angle parameter between the normal vectors; The processing module is used to use the first segmentation parameter and the second segmentation parameter as point cloud segmentation parameters for the masonry body, and to perform point cloud segmentation on the point cloud data of the masonry body to be processed according to the point cloud segmentation parameters.

[0015] Optionally, the determining module is specifically used to determine the curvature information of each point in each sample region based on the point cloud data of each sample region; The curvature information of each point in each sample region is used as the curvature information of the masonry body to be processed. Multiple sub-regions are determined from each sample region, and multiple pairs of points are determined for each sub-region; Based on multiple pairs of points corresponding to each sub-region in each sample region, determine the mean and standard deviation of the angle between the normal vectors; Candidate regions are selected from each sample region, and the set of normal vectors corresponding to each candidate region is determined. Based on the set of normal vectors corresponding to each candidate region in each sample region, determine the reference value of the angle between the normal vectors; The mean and standard deviation of the included angle of the normal vectors, as well as the reference value of the included angle of the normal vectors, are used as the included angle parameters of the normal vectors corresponding to the masonry body to be processed.

[0016] Optionally, the determining module is specifically used to determine the position of the first curvature peak and the position of the second curvature peak based on the curvature information of each point in each sample region; The target valley location is determined based on the positions of the first and second curvature peaks. The first segmentation parameter is determined based on the curvature information of the target valley bottom location and a preset safety factor.

[0017] Optionally, the determining module is specifically used to determine a first threshold based on the mean and standard deviation of the angle between the normal vectors; Determine the second threshold based on the reference value of the angle between the normal vectors; Based on the first threshold and the second threshold, determine the value range of the second segmentation parameter; The second segmentation parameter is determined within the range of values ​​of the second segmentation parameter.

[0018] Optionally, the determining module is specifically used to determine multiple initial sub-regions from the sample region; Based on the number of points contained in each initial sub-region and the curvature information of each point, the initial sub-regions are first filtered to obtain each first-filtered sub-region. Principal component analysis is performed on each point in each first screening sub-region to determine the point set distribution characteristics of each first screening sub-region; Based on the point set distribution characteristics of each first screening sub-region, a second screening is performed on each first screening sub-region to obtain each second screening sub-region; Based on each of the second-selection sub-regions, a third selection is performed to obtain multiple sub-regions corresponding to the sample region.

[0019] Optionally, the determining module is specifically used to determine the normal vector angle information of each pair of points based on multiple pairs of points corresponding to each sub-region in each sample region, so as to obtain the set of normal vector angles corresponding to each sample region. Based on the information of the angles of the normal vectors in the set of angles of the normal vectors corresponding to each sample region, the mean and standard deviation of the angles of the normal vectors are determined.

[0020] Optionally, the determining module is specifically used to perform cluster analysis based on the center point information of the candidate region to determine each cluster corresponding to the candidate region; Based on the distribution information and curvature information of each point in each cluster, the target sub-region in each cluster is determined; Based on the normal vector information of each point within the target sub-region in each cluster, the normal vector corresponding to each cluster is determined, and the normal vector corresponding to each cluster is used as the set of normal vectors corresponding to the candidate region.

[0021] Optionally, the determining module is specifically used to determine the absolute value of the angle between the normal vectors of each pair of clusters based on the normal vectors corresponding to each cluster of each candidate region in the sample region, so as to obtain the set of angles between the normal vectors within the region corresponding to the sample region. Based on the set of angles between the normal vectors within each sample region, a reference value for the angle between the normal vectors is determined.

[0022] Thirdly, embodiments of this application provide an electronic device, including: a processor, a storage medium, and a bus. The storage medium stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to implement the adaptive determination method for point cloud segmentation threshold parameters of masonry as provided in the first aspect.

[0023] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the adaptive determination method for point cloud segmentation threshold parameters of masonry as provided in the first aspect.

[0024] The beneficial effects of this application are: This application provides an adaptive method and apparatus for determining point cloud segmentation threshold parameters for masonry structures. The method includes: determining at least one sample region from the point cloud data of the masonry structure to be processed; determining the curvature information and normal vector angle parameters corresponding to the masonry structure to be processed based on the point cloud data of each sample region; determining a first segmentation parameter based on the curvature information; determining a second segmentation parameter based on the normal vector angle parameters; using the first and second segmentation parameters as point cloud segmentation parameters for the masonry structure, and performing point cloud segmentation on the point cloud data of the masonry structure to be processed according to the point cloud segmentation parameters. This method, based on the characteristics of the point cloud data of the masonry structure itself, extracts the curvature information and normal vector angle information from the point cloud data, thereby achieving automated and adaptive determination of the curvature threshold and normal vector angle threshold. This eliminates reliance on manual experience and ensures the objectivity of the determined curvature threshold and normal vector angle threshold, as well as their matching degree with the masonry structure, improving the accuracy and robustness of point cloud segmentation of the masonry structure. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A flowchart illustrating an adaptive method for determining point cloud segmentation threshold parameters for masonry, provided in an embodiment of this application; Figure 2 A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; Figure 3 A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; Figure 4 A schematic diagram of a statistical distribution histogram of curvature values ​​provided in an embodiment of this application; Figure 5 A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; Figure 6 A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; Figure 7 A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; Figure 8A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; Figure 9 A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; Figure 10 A schematic diagram of an adaptive determination device for point cloud segmentation threshold parameters of masonry provided in an embodiment of this application; Figure 11 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0028] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0029] 3D point cloud segmentation technology based on region growing algorithms is an effective means of processing point clouds of regular stone masonry dams and achieving automatic identification of stones and gaps. The performance of this type of algorithm is highly dependent on the setting of two core threshold parameters: curvature threshold (C_threshold) and normal vector angle threshold (Alpha_threshold). The curvature threshold is used to filter seed points and distinguish between flat stone surfaces and uneven gaps; the normal vector angle threshold is used to control the similarity tolerance of region growing and determine the accuracy of the segmentation boundary.

[0030] Currently, determining the aforementioned threshold relies entirely on the prior knowledge of engineers and technicians and manual trial and error. Specifically, operators preset an initial parameter based on their experience and repeatedly adjust and verify it by observing the segmentation effect (e.g., whether the stones are split or different stones are merged). This method has inherent limitations: Highly subjective and with unstable results: Differences in the experience of different experts lead to inconsistent parameter settings, poor reproducibility and consistency of segmentation results, and difficulty in forming a standardized process.

[0031] Inefficient and costly: For each new engineering project or point cloud data with different acquisition precision, time-consuming parameter adjustments are required, which severely restricts the efficiency of construction quality inspection.

[0032] Difficulty in finding the optimal solution and limited accuracy: Manual trial and error cannot find the globally optimal solution from the perspective of data statistics. It often only yields "usable" rather than "optimal" parameters, which leads to the segmentation accuracy failing to reach the theoretical upper limit and affects the accuracy of subsequent calculations of key indicators such as porosity.

[0033] Therefore, developing a method and system that can automatically, quickly, and objectively determine the optimal threshold parameters has become a key bottleneck in promoting the intelligent detection technology of masonry dams from "semi-automation" to "full automation".

[0034] Based on this, this solution provides an adaptive method for determining the point cloud segmentation threshold parameters of masonry dams. It aims to achieve fully automatic, data-driven determination of the curvature threshold and the angle threshold between the normal vectors, completely eliminating reliance on manual experience. Through statistical analysis and physical feature mining, it ensures that the determined parameters are objective, optimal, or suboptimal, thereby maximizing the accuracy and robustness of the segmentation algorithm. This significantly improves the overall efficiency of point cloud segmentation for masonry dams, enabling it to adapt to the needs of different projects and data sources, and providing core technical support for achieving high-quality, batch-based construction quality inspection.

[0035] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0036] Figure 1 This is a flowchart illustrating an adaptive method for determining point cloud segmentation threshold parameters for masonry structures, provided in an embodiment of this application. The execution entity of this method can be a computer device, such as... Figure 1 As shown, the method may include: S101. Determine at least one sample region from the point cloud data of the masonry body to be processed.

[0037] The sample area includes the surface of regular stones, the gaps between stones, and the edges of the stones.

[0038] The masonry body to be processed refers to the masonry body that needs to be segmented into a point cloud. One or more representative sample regions can be automatically or interactively extracted from the overall point cloud data of the masonry body to be processed.

[0039] The masonry to be processed can be constructed from regular stones, which are stones with regular shapes, clear edges and corners, and flat masonry surfaces, including but not limited to coarse stones and blocks. The sample area extracted above must include complete and easily identifiable flat surfaces of regular stones, gaps between stones, and edges of stone corners.

[0040] In one implementation, at least one sample region can be obtained from the masonry body to be processed through interactive extraction. The specific implementation is as follows: The point cloud data of the masonry body to be processed is loaded and visualized in 3D within the user interface; a region selection tool (such as a cuboid, cube, or lasso selection tool) is provided to the user, allowing the user to directly select one or more sample regions on the point cloud data based on visual observation and experience. These sample regions contain complete, regular, flat stone surfaces, clear gaps between stones, and the edges and corners of the stones. The system can then extract all points within the selected area from the point cloud data, forming a sample region for subsequent analysis.

[0041] In another implementation, sample regions can be automatically extracted. Specifically, the point cloud data of the masonry structure to be processed is uniformly sampled or globally calculated to obtain simplified features for each point, such as local point cloud density and approximate curvature. Based on these features, multiple candidate regions with uniform and gently changing features are identified using a sliding window or region proposal algorithm. These regions likely correspond to the smooth surface of the stone. From all candidate regions, the uniformity of their internal features and the feature differences between them and other candidate regions are evaluated, and one or a few regions that best represent the diversity of the overall masonry structure are automatically selected as the final sample region.

[0042] S102. Based on the point cloud data of each sample area, determine the curvature information and the angle parameters of the normal vector corresponding to the masonry body to be processed.

[0043] The curvature information corresponding to the masonry body to be processed can include the curvature information of each point in each sample area. The curvature information can refer to the curvature value, with each point corresponding to a curvature value.

[0044] The parameters of the angle between the normal vectors can include the mean and standard deviation of the normal vectors, as well as a reference value for the angle between the normal vectors.

[0045] S103. Determine the first segmentation parameter based on the curvature information.

[0046] The first segmentation parameter can be calculated by using the curvature information of the masonry body to be processed. The first segmentation parameter can refer to the curvature threshold used for global segmentation, which is used to filter seed points and distinguish between flat stone surfaces and uneven gaps.

[0047] S104. Determine the second segmentation parameter based on the angle parameter between the normal vectors.

[0048] The second segmentation parameter can be calculated using the angle parameter of the normal vector of the masonry to be processed. The second segmentation parameter can refer to the threshold of the angle of the normal vector used for global segmentation. The threshold of the angle of the normal vector is used to control the similarity tolerance of the region growth and determine the accuracy of the segmentation boundary.

[0049] S105. Use the first segmentation parameter and the second segmentation parameter as the point cloud segmentation parameters of the masonry body, and perform point cloud segmentation on the point cloud data of the masonry body to be processed according to the point cloud segmentation parameters.

[0050] The curvature threshold and the normal vector angle threshold can be used as point cloud segmentation parameters for masonry. Thus, when performing point cloud segmentation on the masonry to be processed based on the region growing algorithm, more accurate segmentation can be achieved based on these two segmentation parameters, thereby improving the point cloud segmentation accuracy of the masonry.

[0051] In other words, this solution is based on the overall point cloud data characteristics of the masonry body itself, and generates segmentation parameters suitable for its own point cloud data segmentation in an automated manner, thereby realizing the adaptive determination of segmentation parameters and improving the accuracy of point cloud segmentation of the masonry body.

[0052] In summary, the adaptive determination method for point cloud segmentation threshold parameters of masonry provided in this embodiment includes: determining at least one sample region from the point cloud data of the masonry to be processed; determining the curvature information and normal vector angle parameters corresponding to the masonry to be processed based on the point cloud data of each sample region; determining a first segmentation parameter based on the curvature information; determining a second segmentation parameter based on the normal vector angle parameters; using the first and second segmentation parameters as point cloud segmentation parameters for the masonry, and performing point cloud segmentation on the point cloud data of the masonry to be processed based on the point cloud segmentation parameters. This method can automatically and adaptively determine the curvature threshold and normal vector angle threshold by extracting the curvature information and normal vector angle information of the point cloud data based on the characteristics of the point cloud data of the masonry itself, thus eliminating the dependence on manual experience, while ensuring the objectivity of the determined curvature threshold and normal vector angle threshold and the degree of matching with the masonry, thereby improving the accuracy and robustness of point cloud segmentation of masonry.

[0053] Figure 2A flowchart illustrating another method for adaptively determining point cloud segmentation threshold parameters for masonry provided in this application embodiment; optionally, in step S102, the curvature information and normal vector angle parameters corresponding to the masonry to be processed are determined based on the point cloud data of each sample region, including: S201. Based on the point cloud data of each sample region, determine the curvature information of each point in each sample region.

[0054] Each sample region corresponds to point cloud data, which is a collection of multiple points. For each sample region, the curvature information of each point in the sample region can be determined.

[0055] S202. Use the curvature information of each point in each sample area as the curvature information of the masonry body to be processed.

[0056] The curvature information of the masonry body to be processed can be obtained from the curvature information of each point in each sample region.

[0057] S203. Determine multiple sub-regions from each sample region, and determine multiple pairs of points corresponding to each sub-region.

[0058] Multiple sub-regions can be selected from the sample area through manual or automatic segmentation. These sub-regions can come from the same stone or from different stones within the sample area. Multiple point pairs can be sampled within each sub-region.

[0059] S204. Based on the multiple pairs of points corresponding to each sub-region in each sample region, determine the mean and standard deviation of the angle between the normal vectors.

[0060] For multiple pairs of points in each sample region, the angle between the normal vectors of the pairs can be calculated. Thus, the mean and standard deviation of the angle between the normal vectors of each pair of points in each sample region can be determined.

[0061] S205. Select candidate regions from each sample region and determine the set of normal vectors corresponding to each candidate region.

[0062] Candidate regions can refer to flat areas belonging to different types of stone. By performing cluster analysis on each candidate region, the candidate regions can be divided into multiple clusters, and the representative normal vector of each cluster can be calculated to obtain the set of normal vectors corresponding to each candidate region.

[0063] S206. Determine the reference value of the angle between the normal vectors based on the set of normal vectors corresponding to each candidate region in each sample region.

[0064] By performing a comprehensive calculation based on the set of normal vectors corresponding to each candidate region in each sample region, a reference value for the angle between the normal vectors can be obtained.

[0065] S207. The mean and standard deviation of the angle between the normal vectors, as well as the reference value of the angle between the normal vectors, are used as the angle parameters of the normal vectors corresponding to the masonry body to be processed.

[0066] The parameters of the included angle of the normal vector corresponding to the masonry body to be processed may include: the mean and standard deviation of the included angle of the normal vector, and the reference value of the included angle of the normal vector.

[0067] Figure 3 A flowchart illustrating another method for adaptively determining point cloud segmentation threshold parameters for masonry provided in this application embodiment; optionally, in step S103, determining the first segmentation parameter based on curvature information includes: S301. Determine the positions of the first curvature peak and the second curvature peak based on the curvature information of each point in each sample region.

[0068] Optionally, based on the curvature information, i.e. curvature values, of each point in each sample region, a statistical distribution histogram of the curvature values ​​of all points can be plotted.

[0069] Figure 4 This is a schematic diagram of a histogram illustrating the statistical distribution of curvature values, provided in an embodiment of this application. The horizontal axis of the histogram represents the curvature value, and the vertical axis represents the number of points falling within the corresponding curvature interval, i.e., the frequency.

[0070] Identify the bimodal distribution structure in the histogram and determine the positions of the first and second curvature peaks. The first curvature peak can refer to the position of the low curvature peak, which corresponds to the flat surface of the stone. The second curvature peak can refer to the position of the high curvature peak, which corresponds to the gaps and edge areas.

[0071] Combination Figure 4 It can be seen that the first curvature peak can be located near curvature 0.02, and the second curvature peak can be located near curvature 0.15.

[0072] S302. Determine the target valley location based on the positions of the first and second curvature peaks.

[0073] Based on the positions of the first and second curvature peaks, the statistical distribution valley between the two peaks, i.e., the target valley position, can be found and determined.

[0074] In this embodiment, the "valley" can refer to the range of curvature values ​​corresponding to the interval (or "bar") with the lowest frequency between the low curvature peak and the high curvature peak in the curvature distribution histogram. Graphically, it is the lowest point of the "trough" connecting the two peaks.

[0075] The method for determining the target valley bottom position can be as follows: First: After identifying the positions of the first curvature peak and the second curvature peak (assuming their curvature indices are i_low and i_high respectively, and i_low < i_high), traverse all curvature intervals within the interval [i_low, i_high] to find the curvature value C_valley corresponding to the minimum frequency. This C_valley is the valley bottom curvature value, and the position where this valley bottom curvature value is located is used as the target valley bottom position. Second: To overcome the inaccurate valley bottom positioning caused by histogram noise or fluctuations, curve fitting techniques can be used. Use a smoothing spline or a Gaussian mixture model to fit the original histogram distribution to obtain a continuous and smooth probability density function. Then, on this fitted curve, find the point where the first derivative is zero and the second derivative is positive (i.e., the local minimum point in mathematics) between the positions of the first curvature peak and the second curvature peak. The curvature value corresponding to this point is the more stable valley bottom estimated value C_valley.

[0076] Continue as Figure 4 shown, the curvature value corresponding to the determined target valley bottom position can be 0.046 in the figure. It should be noted that the threshold value of 0.046 shown in the figure is the critical value automatically found by the computer through analyzing the statistical distribution characteristics of the point cloud data and can most effectively distinguish stones and gaps.

[0077] S303. Determine the first segmentation parameter according to the curvature information of the target valley bottom position and a preset safety factor.

[0078] In some embodiments, the curvature value of the target valley bottom position can be multiplied by a safety factor k1 to obtain the curvature threshold C_threshold for global segmentation, and this curvature threshold is also the first segmentation parameter.

[0079] Among them, the value range of the safety factor k1 can be 0.8 - 1.2.

[0080] Figure 5 FIG. is a schematic flowchart of another method for adaptively determining the point cloud segmentation threshold parameter of a masonry body provided by an embodiment of the present application; optionally, in step S104, determining the second segmentation parameter according to the normal vector angle parameter includes: S401. Determine the first threshold according to the mean and standard deviation of the normal vector angle.

[0081] Optionally, assuming the mean of the angle between the normal vectors is denoted as μ_intra and the standard deviation of the angle between the normal vectors is denoted as σ_intra, the first threshold can be calculated based on the adjustable coefficient and the mean and standard deviation of the angle between the normal vectors, resulting in the following first threshold: μ_intra + N * σ_intra; where N represents the adjustable coefficient, which is usually 2 or 3 to ensure that the threshold covers most of the normal fluctuations within the same surface.

[0082] S402. Determine the second threshold based on the reference value of the angle between the normal vectors.

[0083] The reference value of the angle between the normal vectors can be represented as θ_inter_min, and the second threshold can refer to θ_inter_min.

[0084] S403. Determine the range of values ​​for the second segmentation parameter based on the first threshold and the second threshold.

[0085] Assuming the second segmentation parameter is denoted as Alpha_threshold, then the second segmentation parameter satisfies the following constraints: Alpha_threshold>μ_intra + N *σ_intra, and Alpha_threshold<θ_inter_min.

[0086] S404. Determine the second segmentation parameter within the range of its values.

[0087] In other words, an angle can be selected as the second segmentation parameter within a range that meets the above conditions. This can involve further trials within that range to select the optimal second segmentation parameter.

[0088] Figure 6 A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; optionally, in step S203, multiple sub-regions are determined from each sample region, including: S501. Determine multiple initial sub-regions from the sample region.

[0089] This approach takes the determination of multiple sub-regions from any single sample region as an example. In some embodiments, N initial sub-regions can be selected manually or through preliminary segmentation.

[0090] S502. Based on the number of points contained in each initial sub-region and the curvature information of each point, perform a first screening on each initial sub-region to obtain each first-screened sub-region.

[0091] Since the selection of N initial sub-regions may be too rough, we can then filter and select regions from the N initial sub-regions to determine M flat and usable sub-regions.

[0092] The selection criteria for the first screening are as follows: ① Point count threshold screening: The number of points contained in the region must be greater than a minimum value (the minimum value N_min can be 300 points) to ensure that there are enough points in the region for statistical sampling and to exclude noise regions that are too small.

[0093] ② Flatness quantification screening: Calculate the mean curvature and standard deviation of curvature for all points within each region. Retain regions where the mean curvature is less than C_avg_max and the standard deviation of curvature is less than C_std_max to ensure that the region is generally very flat with minimal fluctuations.

[0094] The mean curvature C_avg_max and standard deviation of curvature C_std_max can be pre-defined parameters. C_avg_max defines the maximum allowable average curvature of a region as a whole, measuring how flat the defined region is on average, and is used to filter out surfaces that are generally curved. C_std_max defines the upper limit of the allowable range of curvature value fluctuations at points within a region, measuring the roughness or inhomogeneity within the region, and is used to filter out regions with rough surfaces or containing noise / small sharp edges.

[0095] Based on the number of points contained in each initial sub-region and the curvature value of each point, the initial sub-regions can be filtered according to the first filtering strategy described above to remove initial sub-regions that do not meet the conditions and retain the initial sub-regions that meet the first filtering strategy as the first filtered sub-regions.

[0096] S503. Perform principal component analysis on each point in each first screening sub-region to determine the point set distribution characteristics of each first screening sub-region.

[0097] Next, geometric morphology screening is performed on each of the first screening sub-regions to enhance robustness. Principal component analysis can be performed on the points within the first screening sub-regions to calculate their eigenvalues.

[0098] The eigenvalues ​​come from principal component analysis. For all points in a first screening sub-region (a three-dimensional point set), the algorithm calculates the covariance matrix of these points, then performs eigenvalue decomposition on this matrix, and obtains three eigenvalues ​​(λ1, λ2, λ3) and their corresponding three eigenvectors. The eigenvalues ​​are usually sorted from largest to smallest: λ1≥λ2≥λ3≥0.

[0099] The three eigenvalues ​​λ1, λ2, and λ3 describe the distribution of the point set along three mutually perpendicular principal directions, which are the corresponding eigenvectors.

[0100] λ1 (the largest eigenvalue) represents the extent of the distribution of the point set along the first principal direction (the length direction), which is the direction in which the point set is most widely distributed.

[0101] λ2 (the second largest eigenvalue) represents the extent of the distribution of the point set in the second principal direction (width direction), which is perpendicular to the first direction and is the most widely distributed of the remaining directions.

[0102] λ3 (the smallest eigenvalue) represents the extent to which the point set is spread along the third principal direction (thickness direction), which is perpendicular to the first two directions; for an ideal plane, the points should hardly change in this direction.

[0103] S504. Based on the point set distribution characteristics of each first screening sub-region, perform a second screening on each first screening sub-region to obtain each second screening sub-region.

[0104] For an ideal flat region, the first eigenvalue λ1 and the second eigenvalue λ2 should be much larger than the third eigenvalue λ3 (λ3 is close to 0). By setting a flatness index (λ3 / (λ1+λ2+λ3)<0.01), non-planar regions located at edges or corners can be excluded.

[0105] The planarity index λ3 / (λ1+λ2+λ3) calculates how much of the total variance of the point set is contributed by the thinnest direction (thickness direction). If the planarity index is <0.01, the region is retained; if the planarity index is ≥0.01, the region is excluded.

[0106] Principal component analysis was used to further filter each of the first screening sub-regions. The sub-regions that were retained after the screening were called the second screening sub-regions.

[0107] S505. Based on each of the second screening sub-regions, perform a third screening to obtain multiple sub-regions corresponding to the sample region.

[0108] For each of the selected second-selection sub-regions, a final selection can be performed to obtain multiple representative sub-regions.

[0109] In one implementation, the second screening sub-regions can be sorted from high to low "flatness" (average curvature from small to large) by automatic filtering, and the top M (M≥3) sub-regions can be automatically selected as multiple sub-regions corresponding to the sample region.

[0110] In another implementation method, the selected second-filter sub-areas can be highlighted to the user through manual screening. The user can then make a visual judgment and finally confirm or fine-tune the selection.

[0111] Figure 7 This is a flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in an embodiment of this application; optionally, in step S204, the mean and standard deviation of the angle between the normal vectors are determined based on multiple pairs of points corresponding to each sub-region in each sample region, including: S601. Based on the multiple pairs of points corresponding to each sub-region in each sample region, determine the angle information of the normal vectors of each pair of points to obtain the set of angles of the normal vectors corresponding to each sample region.

[0112] For the M sub-regions selected in the sample region, multiple pairs of points can be sampled from each sub-region. The angle between the normal vectors of each pair of points is calculated. Therefore, a set of normal vectors is obtained for each sub-region. The set of normal vectors corresponding to all sub-regions constitutes the set of normal vector angles corresponding to the sample region.

[0113] It is worth noting that the set of normal vector angles for each sample region was obtained in the same way.

[0114] It should be noted that multiple point pairs in each sub-region can be sampled in the following manner, where multiple point pairs require a large amount of random and repeatable sampling.

[0115] 1) Number of samples: For a sub-region, the number of samples can be considered as 1000 pairs or as 20% of the number of points in the region; the more pairs of points are sampled, the more stable the calculated statistics (mean and standard deviation) will be and the more representative they will be of the true characteristics of the region.

[0116] 2) Sampling strategy: Simple random sampling: Randomly select two points with replacement from the set of points in the sub-region to form a point pair. This is the simplest method.

[0117] Random sampling without replacement: For computational efficiency, the point set can be randomly shuffled without replacement and then paired sequentially until the required number is reached.

[0118] The set of angles between the normal vectors of a subregion within a sample region can be calculated as follows: 1) Input: A subregion containing K points.

[0119] 2) Cyclic sampling: Repeat the following steps N times (e.g., N=1000): a. Random point selection: Randomly select two different points P_i and P_j from the K points in this sub-region.

[0120] b. Calculate the angle: Obtain the unit normal vector N_i of point P_i; obtain the unit normal vector N_j of point P_j; calculate the angle θ between the two normal vectors = arccos(|N_i·N_j|) (Note: Only the angle is considered, not the direction, so the absolute value is sufficient).

[0121] c. Record: Put this angle value θ into the set of angles of the corresponding normal vector of the sub-region.

[0122] S602. Based on the information of the angles of the normal vectors in the set of angles of the normal vectors corresponding to each sample region, determine the mean and standard deviation of the angles of the normal vectors.

[0123] After merging the set of normal vector angles across all sample regions, the mean and standard deviation of the normal vector angles can be calculated using the arithmetic mean method.

[0124] Assuming the set of normal vector angles for all sample regions contains N normal vector angles, the mean μ_intra and standard deviation σ_intra of the normal vector angles can be calculated using the following formulas: μ_intra=(θ1+θ2+...+θ_N) / N σ_intra=sqrt([Σ(θ i -μ_intra)²] / (N-1)).

[0125] Figure 8 This is a flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in an embodiment of this application; optionally, in step S205, determining the set of normal vectors corresponding to each candidate region includes: S701. Based on the centroid information of the candidate regions, perform cluster analysis to determine the clusters corresponding to the candidate regions.

[0126] First, a conservative region generation segmentation strategy can be applied to the sample area to obtain flat areas of different stone materials. Then, for all the segmented areas, flat areas with more than 100 points are further filtered out. The purpose is to remove extremely small noise areas and build a candidate region pool. The candidate region pool contains candidate regions, retaining more candidate regions, including those that may not be so flat but belong to different stone materials.

[0127] For each candidate region, cluster analysis can be performed to distinguish different types of stone. Optionally, the coordinates of the center point of each candidate region can be calculated. Based on the principle that the center points of multiple regions on the same stone are closely clustered in space, while the center points of regions on different stones are clearly separated by gaps, clustering algorithms can be used, such as Density-Based Spatial Clustering of Applications with Noise (DBSCAN) or hierarchical clustering, to cluster the candidate regions according to the spatial location of the center points, dividing the candidate regions into K clusters, each cluster representing a physical stone.

[0128] S702. Based on the distribution information and curvature information of each point in each cluster, determine the target sub-region in each cluster.

[0129] For each cluster, identify the region within that cluster with the most points or the highest flatness (lowest average curvature), which is the target sub-region. The target sub-region is considered to be the most representative flat surface of the stone.

[0130] S703. Based on the normal vector information of each point in the target sub-region of each cluster, determine the normal vector corresponding to each cluster, and use the normal vector corresponding to each cluster as the set of normal vectors corresponding to the candidate region.

[0131] Calculate the average of the normal vectors of all points within the target sub-region and normalize it to a unit vector to obtain the normal vectors corresponding to the clusters. Thus, each cluster has a normal vector representing its surface orientation. The normal vectors corresponding to each cluster within the candidate region can form the set of normal vectors for the candidate region.

[0132] Figure 9 A flowchart illustrating another method for adaptively determining the point cloud segmentation threshold parameter of masonry provided in this application embodiment; optionally, in step S206, determining the reference value of the angle between the normal vectors based on the set of normal vectors corresponding to each candidate region in each sample region includes: S801. Based on the normal vectors corresponding to each cluster in each candidate region of the sample region, determine the absolute value of the angle between the normal vectors of each pair of clusters, and obtain the set of angles between the normal vectors within the sample region.

[0133] Taking the calculation of a candidate region in a sample region as an example, all the clusters in the candidate region can be traversed. For any two different clusters \(C_i\) and \(C_j\) (where \(i < j\)), their normal vectors \(V_i\) and \(V_j\) are obtained, and the included angle \(\theta_{ij}=\arccos(|V_i\cdot V_j|)\) between the two normal vectors is calculated, and the absolute value is taken. That is, only the included angle is concerned, and the direction of the normal vector is not concerned.

[0134] Thus, each candidate region can obtain the included angles between the normal vectors of multiple clusters; the included angles between the normal vectors corresponding to all candidate regions in a sample region together constitute the set of intra-region included angles between normal vectors corresponding to the sample region.

[0135] S802. Determine the reference value of the included angle between the normal vectors according to the set of intra-region included angles between normal vectors corresponding to each sample region.

[0136] The sets of intra-region included angles between normal vectors corresponding to all sample regions can be merged to obtain the final target set of included angles between normal vectors.

[0137] In one way, the minimum value of the intersection of the normal vectors can be found from the target set of included angles between normal vectors as the reference value of the included angle between the normal vectors.

[0138] In another implementation, to avoid outliers (such as a very small included angle caused by data noise), the 5% quantile of the smallest included angles among all the included angles between the normal vectors can be taken, or the average value of the two smallest included angles between the normal vectors can be taken as the reference value of the included angle between the normal vectors.

[0139] It is worth noting that in this scheme, when determining the threshold of the included angle between the normal vectors, by sampling a large number of point pairs on the flat surfaces of the same stone material that have been strictly screened, and calculating the included angles between their normal vectors, the mean and standard deviation of the included angles between the normal vectors are statistically obtained, thereby quantifying the normal fluctuation range of the normal vectors within the same surface.

[0140] Since the mean and standard deviation of the included angles between the normal vectors are from the statistical results of the actual point cloud data, therefore: if the point cloud resolution is high, the noise is small, and the internal included angle fluctuation is small, the threshold of the included angle between the normal vectors can be smaller, so that the segmentation is more refined; if the point cloud is rough or has perturbations, and the internal included angle fluctuation is large, the threshold of the included angle between the normal vectors is automatically relaxed, which can avoid mis-splitting; at the same time, the reference value of the included angle between the normal vectors reflects the minimum direction difference between different stone materials in the current scene and also dynamically adjusts with the actual structure.

[0141] On this basis, the value constraint conditions for the threshold of the included angle between the normal vectors are set, which ensures that the obtained threshold of the included angle between the normal vectors can tolerate most of the normal changes caused by surface micro-inequality and acquisition noise, avoids wrongly splitting a complete stone material (over-segmentation), and improves the robustness and stability of the algorithm.

[0142] This design starts from the essence of geometry and physics, making full use of the prior knowledge of the regular arrangement of masonry dams and the clear orientation of stones to determine the threshold of the angle between normal vectors, effectively preventing erroneous growth across stones and ensuring the accuracy of the segmentation boundary.

[0143] This method determines the curvature threshold entirely based on the geometric feature distribution of the point cloud data of the masonry to be processed, without human intervention. Curvature is a key indicator for measuring the local geometric smoothness of a point cloud. Regular stone surfaces are flat with low and concentrated local curvature; gaps and edges show drastic changes with high local curvature. This method uses the valley bottom in the bimodal structure as the classification boundary, essentially searching for an optimal inter-class separation point, finding the critical curvature value that best distinguishes the two types of geometric entities (stone vs. gaps). Therefore, the determined curvature threshold is not only reasonable but also has clear physical meaning and geometric interpretability. This method dynamically determines the curvature threshold that best matches the masonry to be processed by conducting on-site analysis of each specific project, automatically adapting to different point cloud densities, noise levels, and scale variations; it does not require pre-setting global empirical values, thus improving the universality and robustness of the method.

[0144] In summary, the adaptive determination method for point cloud segmentation threshold parameters of masonry provided in this embodiment includes: determining at least one sample region from the point cloud data of the masonry to be processed; determining the curvature information and normal vector angle parameters corresponding to the masonry to be processed based on the point cloud data of each sample region; determining a first segmentation parameter based on the curvature information; determining a second segmentation parameter based on the normal vector angle parameters; using the first and second segmentation parameters as point cloud segmentation parameters for the masonry, and performing point cloud segmentation on the point cloud data of the masonry to be processed based on the point cloud segmentation parameters. This method can automatically and adaptively determine the curvature threshold and normal vector angle threshold by extracting the curvature information and normal vector angle information of the point cloud data based on the characteristics of the point cloud data of the masonry itself, thus eliminating the dependence on manual experience, while ensuring the objectivity of the determined curvature threshold and normal vector angle threshold and the degree of matching with the masonry, thereby improving the accuracy and robustness of point cloud segmentation of masonry.

[0145] The following describes the apparatus, equipment, and storage medium used to implement the adaptive determination method for point cloud segmentation threshold parameters of masonry provided in this application. The specific implementation process and technical effects are described above and will not be repeated below.

[0146] Figure 10This is a schematic diagram of an adaptive point cloud segmentation threshold parameter determination device for masonry structures provided in an embodiment of this application. The function implemented by this device corresponds to the steps executed by the method described above. This device can be understood as the aforementioned server, or the server's processor, or as a component independent of the aforementioned server or processor that implements the functions of this application under server control, such as... Figure 10 As shown, the device may include: a determining module 100 and a processing module 200; The determination module 100 is used to determine at least one sample region from the point cloud data of the masonry body to be processed. The sample region includes the regular stone surface, the gaps between stones and the edges of the stones. The determination module 100 is used to determine the curvature information and the angle parameter of the normal vector corresponding to the masonry body to be processed based on the point cloud data of each sample area. The module 100 is used to determine the first segmentation parameters based on the curvature information; The module 100 is used to determine the second segmentation parameter based on the angle parameter between the normal vectors; The processing module 200 is used to use the first segmentation parameter and the second segmentation parameter as point cloud segmentation parameters for the masonry body, and to perform point cloud segmentation on the point cloud data of the masonry body to be processed according to the point cloud segmentation parameters.

[0147] Optionally, the determining module 100 is specifically used to determine the curvature information of each point in each sample region based on the point cloud data of each sample region; The curvature information of each point in each sample region is used as the curvature information of the masonry body to be processed. Multiple sub-regions are determined from each sample region, and multiple pairs of points are determined for each sub-region; Based on multiple pairs of points corresponding to each sub-region in each sample region, determine the mean and standard deviation of the angle between the normal vectors; Candidate regions are selected from each sample region, and the set of normal vectors corresponding to each candidate region is determined. Based on the set of normal vectors corresponding to each candidate region in each sample region, determine the reference value of the angle between the normal vectors; The mean and standard deviation of the angle between the normal vectors, as well as the reference value of the angle between the normal vectors, are used as the parameters of the angle between the normal vectors corresponding to the masonry body to be processed.

[0148] Optionally, the determining module 100 is specifically used to determine the position of the first curvature peak and the position of the second curvature peak based on the curvature information of each point in each sample region; The location of the target valley bottom is determined based on the positions of the first and second curvature peaks. The first segmentation parameter is determined based on the curvature information of the target valley bottom location and the preset safety factor.

[0149] Optionally, the determining module 100 is specifically used to determine the first threshold based on the mean and standard deviation of the angle between the normal vectors; Determine the second threshold based on the reference value of the angle between the normal vectors; The range of values ​​for the second segmentation parameter is determined based on the first threshold and the second threshold. The second segmentation parameter is determined within the range of its values.

[0150] Optionally, the determining module 100 is specifically used to determine multiple initial sub-regions from the sample region; Based on the number of points contained in each initial sub-region and the curvature information of each point, the initial sub-regions are first filtered to obtain each first-filtered sub-region. Principal component analysis is performed on each point in each first screening sub-region to determine the point set distribution characteristics of each first screening sub-region; Based on the point set distribution characteristics of each first screening sub-region, a second screening is performed on each first screening sub-region to obtain each second screening sub-region; Based on each of the second-selection sub-regions, a third selection is performed to obtain multiple sub-regions corresponding to the sample region.

[0151] Optionally, the determining module 100 is specifically used to determine the normal vector angle information of each pair of points based on multiple pairs of points corresponding to each sub-region in each sample region, so as to obtain the set of normal vector angles corresponding to each sample region. Based on the information of the angles of the normal vectors in the set of angles of the normal vectors corresponding to each sample region, determine the mean and standard deviation of the angles of the normal vectors.

[0152] Optionally, the determining module 100 is specifically used to perform cluster analysis based on the center point information of the candidate region to determine each cluster corresponding to the candidate region; Based on the distribution information and curvature information of each point in each cluster, the target sub-region in each cluster is determined; Based on the normal vector information of each point within the target sub-region in each cluster, the normal vector corresponding to each cluster is determined, and the normal vector corresponding to each cluster is used as the set of normal vectors corresponding to the candidate region.

[0153] Optionally, the determining module 100 is specifically used to determine the absolute value of the angle between the normal vectors of each pair of clusters based on the normal vectors corresponding to each cluster of each candidate region in the sample region, so as to obtain the set of angles between the normal vectors within the region corresponding to the sample region. Based on the set of angles between normal vectors within each sample region, determine the reference value of the angle between the normal vectors.

[0154] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

[0155] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more digital signal processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0156] The modules described above can be connected or communicate with each other via wired or wireless connections. Wired connections can include metal cables, optical fibers, hybrid cables, or any combination thereof. Wireless connections can include connections via LAN, WAN, Bluetooth, ZigBee, or NFC, or any combination thereof. Two or more modules can be combined into a single module, and any module can be divided into two or more units. Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here.

[0157] Figure 11 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The device may be a computing device with data processing capabilities.

[0158] The device includes: processor 801 and storage medium 802.

[0159] Storage medium 802 is used to store programs, and processor 801 calls the programs stored in storage medium 802 to execute the above method embodiments. The specific implementation and technical effects are similar, and will not be described in detail here.

[0160] The storage medium 802 stores program code, which, when executed by the processor 801, causes the processor 801 to perform various steps in the adaptive determination method for point cloud segmentation threshold parameters of masonry according to various exemplary embodiments of this application, as described in the "Exemplary Methods" section above.

[0161] The processor 801 can be a general-purpose processor, such as a central processing unit (CPU), digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0162] Storage medium 802, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. The storage medium can include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type storage medium, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic storage medium, magnetic disk, optical disk, etc. The storage medium is any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto. In the embodiments of this application, storage medium 802 can also be a circuit or any other device capable of implementing storage functions for storing program instructions and / or data.

[0163] Optionally, this application also provides a program product, such as a computer-readable storage medium, including a program that, when executed by a processor, performs the above-described method embodiments.

[0164] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0165] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0166] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units.

[0167] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for adaptively determining a point cloud segmentation threshold parameter of a stone masonry, characterized in that, include: At least one sample region is determined from the point cloud data of the masonry body to be processed, the sample region including the regular stone surface, the gaps between stones and the edges of the stones; Based on the point cloud data of each sample area, determine the curvature information and the angle parameter of the normal vector corresponding to the masonry body to be processed; Based on the curvature information, the first segmentation parameter is determined; The second segmentation parameter is determined based on the angle parameter between the normal vectors. The first segmentation parameter and the second segmentation parameter are used as point cloud segmentation parameters for the masonry body, and the point cloud data of the masonry body to be processed is segmented according to the point cloud segmentation parameters.

2. The method of claim 1, wherein, The step of determining the curvature information and normal vector angle parameters of the masonry body to be processed based on the point cloud data of each sample region includes: Based on the point cloud data of each sample region, determine the curvature information of each point in each sample region; The curvature information of each point in each sample region is used as the curvature information of the masonry body to be processed. Multiple sub-regions are determined from each sample region, and multiple pairs of points are determined for each sub-region; Based on multiple pairs of points corresponding to each sub-region in each sample region, determine the mean and standard deviation of the angle between the normal vectors; Candidate regions are selected from each sample region, and the set of normal vectors corresponding to each candidate region is determined. Based on the set of normal vectors corresponding to each candidate region in each sample region, determine the reference value of the angle between the normal vectors; The mean and standard deviation of the included angle of the normal vectors, as well as the reference value of the included angle of the normal vectors, are used as the included angle parameters of the normal vectors corresponding to the masonry body to be processed.

3. The method of claim 2, wherein, Determining the first segmentation parameter based on the curvature information includes: Based on the curvature information of each point in each sample region, determine the positions of the first curvature peak and the second curvature peak; The target valley location is determined based on the positions of the first and second curvature peaks. The first segmentation parameter is determined based on the curvature information of the target valley bottom location and a preset safety factor.

4. The method of claim 2, wherein, The step of determining the second segmentation parameter based on the angle parameter of the normal vector includes: The first threshold is determined based on the mean and standard deviation of the angle between the normal vectors. Determine the second threshold based on the reference value of the angle between the normal vectors; Based on the first threshold and the second threshold, determine the value range of the second segmentation parameter; The second segmentation parameter is determined within the range of values ​​of the second segmentation parameter.

5. The method according to claim 2, characterized in that, The process of determining multiple sub-regions from each sample region includes: Multiple initial sub-regions are determined from the sample region; Based on the number of points contained in each initial sub-region and the curvature information of each point, the initial sub-regions are first filtered to obtain each first-filtered sub-region. Principal component analysis is performed on each point in each first screening sub-region to determine the point set distribution characteristics of each first screening sub-region; Based on the point set distribution characteristics of each first screening sub-region, a second screening is performed on each first screening sub-region to obtain each second screening sub-region; Based on each of the second-selection sub-regions, a third selection is performed to obtain multiple sub-regions corresponding to the sample region.

6. The method according to claim 2, characterized in that, The step of determining the mean and standard deviation of the angle between normal vectors based on multiple pairs of points corresponding to each sub-region in each sample region includes: Based on multiple pairs of points corresponding to each sub-region in each sample region, the angle information of the normal vectors of each pair of points is determined, and the set of angles of the normal vectors corresponding to each sample region is obtained. Based on the information of the angles of the normal vectors in the set of angles of the normal vectors corresponding to each sample region, the mean and standard deviation of the angles of the normal vectors are determined.

7. The method according to claim 2, characterized in that, The determination of the set of normal vectors corresponding to each candidate region includes: Based on the centroid information of the candidate regions, cluster analysis is performed to determine the clusters corresponding to the candidate regions; Based on the distribution information and curvature information of each point in each cluster, the target sub-region in each cluster is determined; Based on the normal vector information of each point within the target sub-region in each cluster, the normal vector corresponding to each cluster is determined, and the normal vector corresponding to each cluster is used as the set of normal vectors corresponding to the candidate region.

8. The method according to claim 7, characterized in that, The step of determining the reference value of the angle between normal vectors based on the set of normal vectors corresponding to each candidate region in each sample region includes: Based on the normal vectors corresponding to each cluster in each candidate region of the sample region, the absolute value of the angle between the normal vectors of each pair of clusters is determined, and the set of angles between the normal vectors within the sample region is obtained. Based on the set of angles between the normal vectors within each sample region, a reference value for the angle between the normal vectors is determined.

9. An adaptive determination device for point cloud segmentation threshold parameters of masonry, characterized in that, include: Determine the module and the processing module; The determining module is used to determine at least one sample region from the point cloud data of the masonry body to be processed. The sample region includes the regular stone surface, the gaps between stones, and the edges of the stones. The determining module is used to determine the curvature information and normal vector angle parameters of the masonry body to be processed based on the point cloud data of each sample area. The determining module is used to determine the first segmentation parameter based on the curvature information; The determining module is used to determine the second segmentation parameter based on the angle parameter between the normal vectors; The processing module is used to use the first segmentation parameter and the second segmentation parameter as point cloud segmentation parameters for the masonry body, and to perform point cloud segmentation on the point cloud data of the masonry body to be processed according to the point cloud segmentation parameters.

10. The apparatus according to claim 9, characterized in that, The determining module is specifically used to determine the curvature information of each point in each sample region based on the point cloud data of each sample region. The curvature information of each point in each sample region is used as the curvature information of the masonry body to be processed. Multiple sub-regions are determined from each sample region, and multiple pairs of points are determined for each sub-region; Based on multiple pairs of points corresponding to each sub-region in each sample region, determine the mean and standard deviation of the angle between the normal vectors; Candidate regions are selected from each sample region, and the set of normal vectors corresponding to each candidate region is determined. Based on the set of normal vectors corresponding to each candidate region in each sample region, determine the reference value of the angle between the normal vectors; The mean and standard deviation of the included angle of the normal vectors, as well as the reference value of the included angle of the normal vectors, are used as the included angle parameters of the normal vectors corresponding to the masonry body to be processed.

Images