A monument three-dimensional monitoring data processing method based on aerial images

Abstract

The application relates to the technical field of image processing, in particular to a historical site three-dimensional monitoring data processing method based on aerial images. The method comprises the following steps: collecting three-dimensional point cloud data of a historical site; processing the three-dimensional point cloud data by using a clustering algorithm to obtain a plurality of target clustering clusters; the plurality of target clustering clusters correspond to a plurality of structure regions of the historical site; calculating the damage score of all corner points in the structure region by calculating the angle difference value of the normal vectors of the fitting surfaces where the corner points and their symmetric points are located; and finally evaluating the overall damage of the historical site according to the damage proportion at the corner points. The damage score is determined by calculating the angle difference value of the normal vectors of the fitting surfaces where the corner points and their symmetric points are located, the symmetry characteristics of the historical site structure are effectively utilized, the subtle damage is converted into quantitative data, the structural deformation caused by natural or human factors can be sensitively captured, the misjudgment of the complex curvature region and the damage region is avoided, and the accuracy and reliability of damage identification are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and particularly relates to a historical site three-dimensional monitoring data processing method based on aerial images. BACKGROUND

[0002] As an important carrier of human history and culture, historical sites carry rich historical information and cultural value. However, in the long years, historical sites have been continuously eroded by natural environment and damaged by human activities. Many historical sites have cracks, peeling, collapse and other damages of different degrees. These damages not only affect the appearance and structural stability of historical sites, but also may lead to the loss of their historical and cultural value.

[0003] With the development of aerial photogrammetry and three-dimensional reconstruction technology, the acquisition of historical site three-dimensional point cloud data provides strong data support for in-depth analysis of the damage of historical sites. Therefore, through the analysis of three-dimensional point cloud data, the damage degree and position of historical sites can be more intuitively and accurately understood, and scientific basis can be provided for protection and repair decisions.

[0004] However, in practical applications, accurate identification of historical site damage areas faces many challenges. On the one hand, after historical sites are damaged by natural disasters or human activities, complex and irregular deformation damage occurs on the surface. On the other hand, the building structure of historical sites has obvious curvature changes in areas such as eaves, angle columns and carved columns. These areas and damage areas have similar geometric features in three-dimensional point cloud data, making it difficult for traditional three-dimensional point cloud-based damage identification methods to effectively distinguish between damage areas and normal curvature change areas, which may easily cause damage misjudgment or omission. This not only leads to unnecessary repair cost waste, but also may delay the repair time of the real damage part due to the wrong evaluation, and further affect the accuracy and effectiveness of historical site protection and repair work. SUMMARY

[0005] To solve the technical problem that it is difficult to effectively distinguish between damage areas and normal curvature change areas in historical site monitoring, the present application provides a historical site three-dimensional monitoring data processing method based on aerial images, which comprises:

[0006] Collecting three-dimensional point cloud data of the historical site; processing the three-dimensional point cloud data using a clustering algorithm to obtain a plurality of target clustering clusters; the plurality of target clustering clusters correspond to a plurality of structural regions of the historical site; calculating a damage score at all corner points in any structural region, including: obtaining any corner point of the current structural region and its symmetric point, when the symmetric point is another corner point of the structural region, determining a first fitting surface according to the data points of the corner point and its neighborhood window, determining a second fitting surface according to the data points of the symmetric point and its neighborhood window, calculating the difference between the included angle of the normal vector of any point pair on the first fitting surface and the included angle of the normal vector of the symmetric point pair on the second fitting surface, and marking it as an angle difference value, and the damage score is equal to the sum of the angle difference values of all point pairs on the first fitting surface; evaluating the damage of the historical site according to the damage scores at all corner points of the plurality of structural regions.

[0007] The present application collects three-dimensional point cloud data and clusters and divides structural regions to realize accurate deconstruction of historical site structures. By calculating the difference in the included angle of the normal vectors of the fitting surfaces of the corner points and their symmetric points, the damage score is determined, which effectively utilizes the symmetry characteristics of the historical site structures and converts subtle damage into quantitative data. Compared with traditional methods, this method can sensitively capture structural deformation caused by natural or human factors, avoid misjudgment in complex curvature regions and damaged regions, and improve the accuracy and reliability of damage identification. Finally, the overall damage of the historical site is evaluated by comprehensively considering the damage scores of the corner points of each structural region, which provides scientific and accurate data support for historical site protection and repair, and improves the efficiency of historical site monitoring and the scientificity of protection decisions.

[0008] As a further improvement of the method of the present application, the angle difference value ; wherein, is the cosine similarity of the normal vector of the point pair on the first fitting surface corresponding to the i-th corner point in the j-th structural region, is the cosine similarity of the normal vector of the symmetric point pair on the second fitting surface, is the inverse cosine function. Accurately measuring the difference in the included angle of the normal vectors of the fitting surfaces at the corner points of the historical site, taking the cosine similarity as the quantitative basis, quantifying the damage as a numerical value, and effectively capturing the changes in the surface caused by damage. It can more accurately distinguish between damaged and normal regions, reduce misjudgment and omission, and provide high-precision quantitative indicators for historical site damage assessment, improving the scientificity and accuracy of historical site damage identification.

[0009]

[0010] ​​​As a further improvement of the method of the present application, the obtaining of any corner point of the current structure region and its symmetric point comprises: determining the coordinate point of any corner point of the structure region by a corner point detection algorithm; calculating the coordinate point of the symmetric point of the corner point ; wherein, is the coordinate point of the symmetric point of the th corner point in the th structure region, is the coordinate point of the center point of the th structure region, is the coordinate point of the th corner point in the th structure region.

[0011] The corner point is accurately positioned by the corner point detection algorithm, providing a reliable basis for subsequent analysis. The position of the symmetric point is calculated using a specific formula, and the position of the symmetric point is quickly and accurately solved based on the center point of the structure region. This process improves the efficiency and accuracy of obtaining the position of the corner point and its symmetric point, provides accurate data for subsequent operations such as calculating the damage score based on the corner point and the symmetric point, and thus improves the accuracy and reliability of the ancient monument damage assessment.

[0012] As a further improvement of the method of the present application, when the symmetric point is not another corner point of the structure region, the damage score is set to 1.

[0013] This way simplifies special case processing, quickly determines the damage score in the atypical symmetric point scenario, avoids complex calculations, improves damage assessment efficiency, and ensures the coherence of the assessment process.

[0014] As a further improvement of the method of the present application, the processing of the three-dimensional point cloud data using a clustering algorithm to obtain a plurality of target clustering clusters comprises: constructing a plurality of initial clustering clusters based on the three-dimensional point cloud data; applying a hierarchical clustering algorithm to the plurality of initial clustering clusters to obtain a plurality of basic clustering clusters; and assigning data points in the three-dimensional point cloud data that are not selected as initial clustering clusters to corresponding basic clustering clusters to obtain target clustering clusters.

[0015] This invention, in the processing of 3D monitoring data for historical sites, effectively aggregates similar data by constructing initial clusters and applying an agglomerative hierarchical clustering algorithm, reducing the complexity of 3D point cloud data and highlighting the structural features of historical sites. Subsequent allocation of remaining data points precisely divides the target clusters, achieving detailed differentiation of structural regions of historical sites, significantly improving data processing efficiency, and laying a precise and orderly data foundation for damage assessment. Simultaneously, addressing the issue of poor clustering results at intersections of different structural regions due to differences in point cloud features, this invention analyzes the differences between point cloud data and their local features, eliminating initial clusters that affect aggregation, further improving the accuracy of segmentation, ensuring that the clustering results more closely match the actual structure of the historical site, and significantly enhancing the scientific rigor and accuracy of historical site damage assessment.

[0016] As another improvement to the method of the present invention, the step of constructing multiple initial clusters based on the three-dimensional point cloud data includes: calculating the gray-level feature similarity between any data point in the point cloud data and its neighboring data points; when the gray-level feature similarity of the data point is greater than its preset similarity threshold, the data point is taken as an initial cluster; the gray-level feature similarity... ;in, It is the first point cloud data Similarity of grayscale features of data points; It is the first The grayscale value of each data point; It is the first Within the neighborhood of the data point, the first The grayscale value of each data point; It is the number of data points in the neighborhood; It is the absolute value symbol; It is a natural exponential function. It is a maximum value function.

[0017] This invention constructs initial clusters by calculating the similarity of grayscale features between point cloud data points and their neighboring points, effectively selecting data points with similar grayscale features as the starting point for clustering. This method uses a specific formula to accurately quantify similarity, avoiding the blindness of randomly selecting initial clusters. Determining the initial cluster when the similarity exceeds a threshold allows the clustering to better reflect the inherent distribution characteristics of the data, improving the rationality and scientific rigor of the initial cluster construction. This lays a solid foundation for subsequent operations such as agglomerative hierarchical clustering, thereby enhancing the accuracy and reliability of clustering and classifying 3D point cloud data of historical sites.

[0018] As another improvement to the method of the present invention, the step of allocating data points in the three-dimensional point cloud data that were not selected as initial clusters to corresponding basic clusters includes: calculating the Euclidean distance from any of the data points that were not selected as initial clusters to the centers of all basic clusters; and allocating the data points that were not selected as initial clusters to the nearest basic cluster.

[0019] By calculating the Euclidean distance from unselected data points to the centers of the basic clusters and assigning them to the nearest basic clusters, the clustering results can be improved in a simple and efficient way. This distance-based assignment strategy makes the data points more reasonably assigned, enhances the accuracy and completeness of clustering, helps to more accurately delineate the structural areas of ancient sites, and provides an ordered data foundation for subsequent damage analysis.

[0020] As a further improvement to the method of the present invention, the step of assessing the damage status of the historical site based on the damage scores at all corner points of the plurality of structural regions includes: marking the corner point when the damage score at the corner point is greater than a preset acceptable damage level; and marking the historical site as severely damaged when the ratio of the number of marked corner points to the total number of corner points of the historical site is greater than a preset ratio.

[0021] The damage to historical sites is assessed by comparing the corner damage scores with preset standards. First, corners exceeding the threshold are marked, and then the degree of damage is determined based on the proportion of marked corners. This method quantifies and standardizes damage assessment, enabling a quick and accurate determination of the overall degree of damage to historical sites and providing an intuitive and clear basis for historical site protection decisions.

[0022] As a further improvement to the method of the present invention, the determination of the first fitting surface and the second fitting surface includes: determining the mathematical model of the fitting surface, wherein the mathematical model includes a quadratic surface and a parametric surface; and calculating the parameters of the mathematical model using the least squares method to obtain the fitting surface.

[0023] By selecting mathematical models such as quadratic surfaces and parametric surfaces, and using the least squares method to calculate parameters, the fitted surface is determined. This method provides a scientific means to analyze the surface morphology near the corners of historical sites, accurately fitting the actual surface and laying a reliable geometric analysis foundation for accurately calculating the corner damage score and assessing the damage situation.

[0024] As another improvement to the method of the present invention, the acquisition of three-dimensional point cloud data of the ancient site includes: taking pictures of the ancient site using aerial photography equipment; constructing the three-dimensional structure of the ancient site through the MVS algorithm to obtain the three-dimensional point cloud data of the ancient site.

[0025] By utilizing aerial photography equipment combined with the MVS algorithm to acquire 3D point cloud data of historical sites, spatial information about these sites can be collected quickly and comprehensively. This method can capture images from multiple angles in the air and construct 3D structures, effectively covering the entire landscape of the historical site and obtaining high-precision point cloud data. This provides a rich and accurate data source for subsequent processing such as damage analysis of historical sites.

[0026] The application fully considers the design characteristics and symmetry of historical buildings, accurately calculates the position of the symmetry point of each corner point by means of a specific algorithm and formula. By analyzing the angle difference of the normal vector direction of the corner point near point cloud and its symmetry point near point cloud on the fitted surface, the damage possibility at each corner point is quantified. This way can sensitively capture the subtle changes caused by damage, effectively distinguish actual damage from normal structure variation, avoid misjudgment and omission, and greatly improve the accuracy of historical damage analysis. At the same time, the local gray feature similarity index is introduced, and based on strict calculation and screening mechanism, high-quality initial clusters are selected. This operation can avoid the adverse interference of low-similarity point clouds in the transition area on the clustering process, optimize the performance of the traditional AGNES algorithm in the division of complex historical structures, and significantly improve the accuracy of clustering, providing a more reliable data basis for subsequent historical damage assessment based on clustering results. BRIEF DESCRIPTION OF DRAWINGS

[0027] Figure 1 A flowchart of a historical relic three-dimensional monitoring data processing method based on aerial images is provided for the embodiments of the application. DETAILED DESCRIPTION

[0028] The embodiments of the application provide a historical relic three-dimensional monitoring data processing method based on aerial images, as shown in the figure, the method comprises steps S100-S400: Figure 1

[0029] Step S100, collect three-dimensional point cloud data of the historical relic.

[0030] To be specific, aerial equipment needs to be used to collect images of the historical building first. In order to establish the three-dimensional structure of the historical building, the aerial equipment needs to be used to take or scan the historical building from multiple angles when collecting images, so as to ensure that the details and structural information of the surface of the historical building can be obtained.

[0031] The MVS algorithm can recover the three-dimensional geometric information of the scene from multiple images with overlapping areas. In the application, the MVS algorithm is used to construct the three-dimensional structure of the historical relic from the collected multi-angle images, and the MVS algorithm will output the three-dimensional point cloud data of the historical relic.

[0032] Step S200, process the three-dimensional point cloud data using a clustering algorithm to obtain a plurality of target clustering clusters, and the plurality of target clustering clusters correspond to a plurality of structural regions of the historical relic.

[0033] ​To elaborate, the structure of historical buildings often comprises multiple elements, such as doors, windows, eaves, and columns, each with significantly different geometric and grayscale characteristics. Analyzing the entire three-dimensional structure of a historical building can lead to interference between these different structural features, resulting in inaccurate damage identification. Therefore, it is necessary to classify the different structures of historical buildings. In practical applications, clustering algorithms can be used to classify these different structures.

[0034] Taking the agglomerative hierarchical clustering algorithm as an example, this algorithm treats each feature point as an initial cluster and iteratively aggregates them to obtain the target cluster. However, for intersecting regions of different structural areas, since the point cloud feature points in the intersecting regions differ from those in other local point cloud feature points, using these point cloud feature points as initial clusters for aggregation results in poor-quality basic clusters. Continuing iterative aggregation will further propagate low-similarity features to the target cluster obtained after aggregation, leading to poor final clustering results that cannot accurately correspond to the actual different structural regions of the historical site. Therefore, this invention improves the clustering algorithm for dividing the structural regions of historical sites.

[0035] Specifically, first, calculate the grayscale similarity between any data point in the point cloud data and its neighboring data points; grayscale similarity:

[0036] ;

[0037] in, It is the first point cloud data Similarity of grayscale features of data points; It is the first The grayscale value of each data point; It is the first Within the neighborhood of the data point, the first The grayscale value of each data point; It is the number of data points in the neighborhood; It is the absolute value symbol; It is a natural exponential function. It is a maximum value function.

[0038] In this formula, the neighborhood refers to a region centered on the data point with a size of [missing information]. The window size can be set according to the condition and needs of the historical building during actual use. The value represents the difference of the current three-dimensional point cloud and the point cloud in its neighborhood in the gray feature. The smaller the value, the closer the corresponding gray value of the local point cloud of the current three-dimensional point cloud, that is, the more similar the local gray feature of the current three-dimensional point cloud, which means that the local point cloud of the current three-dimensional point cloud is more likely to be distributed in the same structure area of the ancient monument. On the contrary, the difference of the corresponding gray value of the local point cloud of the current three-dimensional point cloud is more significant, that is, the local gray feature of the current three-dimensional point cloud is mixed, which means that the current three-dimensional point cloud is likely to be located in a structure transition area.

[0039] Secondly, when the gray feature similarity of the data point is greater than the preset similarity threshold, the data point is selected as an initial clustering cluster.

[0040] In general, there are two or more structure transition areas in ancient architectural buildings, and the more mixed the local gray feature of the three-dimensional point cloud in the transition area. If the feature points of the three-dimensional point cloud in this structure area are selected as initial clustering clusters, the subsequent clustering process will be misled, and different structures will be mixed together, reducing the clustering accuracy.

[0041] Therefore, in order to prevent the three-dimensional point cloud data points in the structure transition area from being selected as initial clustering clusters, the gray feature similarity of each data point is calculated, and the data points greater than the similarity threshold are selected as initial clustering clusters by setting the similarity threshold. In this way, it can be ensured that the initial clustering clusters are not in the structure transition area. The threshold value can be set to 0.6, or it can be set according to requirements.

[0042] Then, the agglomerative hierarchical clustering algorithm is applied to the plurality of initial clustering clusters to obtain a plurality of basic clustering clusters.

[0043] After the initial cluster screening is completed, the traditional agglomerative hierarchical clustering algorithm such as the AGNES algorithm is applied to the initial cluster, which can combine a plurality of initial clustering clusters to obtain a plurality of basic clustering clusters.

[0044] Finally, the data points in the three-dimensional point cloud data that are not selected as initial clustering clusters are distributed to the corresponding basic clustering clusters to obtain target clustering clusters.

[0045] After the above operations are completed, there are still data points with gray feature similarity less than the similarity threshold, which need to be distributed to the basic clustering clusters. In this way, all data points in the three-dimensional point cloud data have corresponding clustering clusters, that is, a plurality of target clustering clusters are obtained according to the three-dimensional point cloud data. According to the above clustering, a plurality of target clustering clusters correspond to a plurality of structure areas of the ancient architectural building.

[0046] The specific data point distribution rule is that for each three-dimensional point cloud data point that is not selected as an initial clustering cluster, the Euclidean distance of the data point to the center of all basic clustering clusters is obtained, the nearest basic clustering cluster is selected for merging, and each data point is distributed to the corresponding basic clustering cluster. The distribution rule can be adjusted according to actual requirements, and is not limited to the nearest distance.

[0047] Step S300: Calculate the damage score at all corner points within any structural region.

[0048] In more detail, a corner point refers to a point where the curvature changes significantly. When historical buildings are affected by natural disasters or human damage, their surface structures suffer varying degrees of damage. The damaged area exhibits significant deformation compared to the surrounding normal area, with a marked change in curvature. Historical buildings often contain areas with different curved contours, such as arches and fan-shaped windows, where the curvature changes significantly. Therefore, calculating damage solely based on corner points can lead to misidentification of the damaged area.

[0049] For ancient buildings, the construction process takes into account factors such as mechanical stability and aesthetics. Each structure has a certain degree of symmetry, which means that the symmetrical position of the corner point of the normal building outline within the same structural area is also a corner point, thus providing a reliable basis for distinguishing damaged areas.

[0050] Therefore, to calculate the degree of damage to a structural region, we can start from the corner points and their symmetrical points. For any structural region, the corner points and symmetrical points are obtained as follows:

[0051] Specifically, firstly, corner detection algorithms such as Harris corner detection, Shi-Tomasi corner detection, and FAST feature detection are used to obtain all corners and their coordinates within the structural region. Then, the center point of the structural region is calculated using the average of the coordinates of all 3D point cloud data within the region. Finally, the coordinates of the symmetrical points are calculated.

[0052] ;

[0053] in, It is the first The first structural region The coordinates of the points symmetrical to the corner points It is the first The coordinates of the center point of each structural region It is the first The first structural region The coordinates of each corner point. It should be noted that all points mentioned in this formula are three-dimensional coordinate points.

[0054] According to the property of symmetry, the symmetrical point and the corner point are symmetrical about the center point, that is, the center point is equal to the average position of the corner point and the symmetrical point. This reflects the fact that the coordinates of the symmetrical point are obtained by extending the vector in the opposite direction of the current corner point relative to the center point.

[0055] The previous section explained how to obtain structural regions, corner points, and symmetrical points. The following section explains how to calculate the damage score at the corner points.

[0056] Regarding the calculation basis of the damage score, the distribution of point clouds typically reflects the geometry of the structural region, and the direction of the normal vector of the fitted surface of the local point cloud is closely related to the distribution of the point cloud. If the point cloud exhibits significant curvature in a certain direction, then the direction of the normal vector of the fitted surface will be related to the direction of curvature. The symmetrical structure of a building means that its contour curves have similar geometric characteristics at relatively symmetrical positions during design, which also means that the overall change in the direction angle of the point cloud normal vector at two symmetrical positions is relatively consistent. Once the overall direction angle of the two symmetrical positions differs significantly, it may indicate that the current area has been affected by external forces, resulting in local changes in geometry (such as curvature or deformation), and the current reference point is considered to be more likely to be damaged.

[0057] Therefore, to calculate the damage score, it is necessary to construct fitting surfaces for each corner point and its symmetrical point. A fitting surface is a mathematically defined surface that best approximates or passes through a set of 3D data points. The construction method for the fitting surface is as follows: First, the data points are determined. This invention uses corner points or symmetrical points and multiple points closest to the corner points or symmetrical points as data points for fitting. Five points are preset, but can be adjusted according to actual needs. Next, the mathematical model of the fitting surface, such as a quadratic surface or a parametric surface, is determined. Finally, the parameters of the mathematical model are calculated using the least squares method to obtain the fitting surface. The determination of the fitting surface is existing technology and will not be elaborated upon here.

[0058] The specific formula for calculating the damage score is as follows:

[0059] ;

[0060] in, ;

[0061] in, Indicates the first The first structural region Points were scored for damage at each corner. Represents the standard normalized function. Indicates the first The first structural region The degree of change in the angle of the normal vector direction within the neighborhood of each corner point and its symmetrical point. Indicates the first The first structural region The coordinates of the points symmetrical to the corner points Indicates the first The coordinates of all corner points of each structural region It is the inverse cosine function. an index value representing two data points in a point pair on the first fitted surface, an index value representing the total number of data points of the first fitted surface or the second fitted surface, is the cosine similarity of the normal vector of a point pair on the first fitted surface corresponding to the i-th corner point in the j-th structural region, is the cosine similarity of the normal vector of a point pair on the first fitted surface corresponding to the i-th corner point in the j-th structural region, is the cosine similarity of the normal vector of a point pair on the first fitted surface corresponding to the i-th corner point in the j-th structural region, is the cosine similarity of the normal vector of a point pair on the first fitted surface corresponding to the i-th corner point in the j-th structural region, is the cosine similarity of the normal vector of a point pair on the first fitted surface corresponding to the i-th corner point in the j-th structural region, is the cosine similarity of the normal vector of a point pair on the first fitted surface corresponding to the i-th corner point in the j-th structural region, represents the angle difference value.

[0062] It should be noted that the point pair is any two points on the first fitted surface, and the symmetric point pair is the two points on the second fitted surface that are symmetric to the point pair on the first fitted surface.

[0063] is the difference between the angle of the normal vector of the two points on the first fitted surface and the angle of the normal vector of the two points on the second fitted surface.

[0064] represents that the symmetric point is not another corner point of the structural region, indicating that the symmetry of the ancient monument structure is destroyed, directly determining that the structural region where the current corner point is located is a damaged region, and setting the damage score to 1. represents that the symmetric point is another corner point of the structural region, reflects the sum of the direction angle differences of the corresponding normal vectors of any two points in the point cloud data of the fitted surface corresponding to the current corner point and its symmetric point, the greater the value, the worse the consistency of the direction angles of the normal vectors in the neighborhood of the current corner point and its symmetric point, which means that the symmetry of the current corner point is destroyed and there is a significant difference in bending in the local area, indicating that the current corner point is more likely to be damaged.

[0065] The above calculations are repeated to obtain the damage score of each corner point in the region.

[0066] Step S400, according to the damage scores of all corner points of the plurality of structural regions, the damage condition of the ancient monument is evaluated.

[0067] In detail, according to the above calculation method, the damage scores of all corner points of all structural regions can be calculated, and a reasonable damage acceptance degree is set according to actual experience, and in this example, the experience value is 0.7. If the damage score of the current corner point corresponding to the ancient monument of different structural regions is greater than the damage acceptance degree, it indicates that the structural region where the current corner point is located has a high possibility of damage, and may have occurred structural damage or deformation, and the current corner point is marked.

[0068] If the total number of the current historical building marker corner points and the total corner point number ratio of the historical building exceeds the preset ratio 5%, it is considered that the current historical building has a more serious damage condition, which may have affected the overall structural stability of the historical building, and immediately sends a warning to inform the cultural relic protection personnel to repair in time. The preset ratio can be set according to the historical relic situation and actual demand.

[0069] The above are preferred embodiments of the present application, not to limit the protection scope of the present application, therefore: any equivalent changes made according to the structure, shape, principle of the present application should be covered within the protection scope of the present application.

Claims

1. A method for processing three-dimensional monitoring data of historical sites based on aerial images, characterized by, The method comprises the following steps: collecting three-dimensional point cloud data of the historical site; processing the three-dimensional point cloud data using a clustering algorithm to obtain a plurality of target clustering clusters; the plurality of target clustering clusters correspond to a plurality of structural regions of the historical site; calculating a damage score at any corner point in any structural region, comprising: obtaining any corner point and its symmetric point of the current structural region, when the symmetric point is another corner point of the structural region, determining a first fitting surface according to the data points of the corner point and its neighborhood window, determining a second fitting surface according to the data points of the symmetric point and its neighborhood window, calculating the difference between the included angle of the normal vector of any point pair on the first fitting surface and the included angle of the normal vector of the symmetric point pair on the second fitting surface, and recording it as the angle difference value, and the damage score is equal to the sum of the angle difference values of all point pairs on the first fitting surface; the obtaining any corner point and its symmetric point of the current structural region comprises: determining the coordinate point of any corner point of the structural region by a corner point detection algorithm; calculating a coordinate point of a symmetry point of the corner point ; in, It is the first The first structural region The coordinates of the symmetrical points of the corner points It is the first The coordinates of the center point of each structural region It is the first The first structural region The coordinates of the corner points; when the symmetric point is not another corner point of the structural region, setting the damage score equal to 1; evaluating the damage of the historical site according to the damage scores at all corner points of the plurality of structural regions. 2.The method of claim 1, wherein, the angle difference value ; wherein, is the cosine similarity of the normal vector of the point pair on the first fitting surface corresponding to the th corner point in the th structure region, is the cosine similarity of the normal vector of the point pair on the first fitting surface and the symmetric point pair on the second fitting surface, is the inverse cosine function. 3.The method of claim 1, wherein, the processing of the three-dimensional point cloud data using a clustering algorithm to obtain a plurality of target clustering clusters comprises: constructing a plurality of initial clustering clusters based on the three-dimensional point cloud data; applying a condensation hierarchical clustering algorithm to the plurality of initial clustering clusters to obtain a plurality of basic clustering clusters; assigning the data points in the three-dimensional point cloud data that are not selected as initial clustering clusters to the corresponding basic clustering clusters to obtain target clustering clusters. 4.The method of claim 3, wherein, the construction of a plurality of initial clustering clusters based on the three-dimensional point cloud data comprises: calculating the gray feature similarity of any data point in the point cloud data with the data points in its neighborhood; when the gray feature similarity of the data point is greater than its preset similarity threshold, the data point is selected as an initial clustering cluster; the gray scale feature similarity ; wherein, is the gray scale feature similarity of the th data point in the point cloud data; is the gray scale value of the th data point; is the gray scale value of the th data point in the neighborhood of the th data point; is the number of data points in the neighborhood; is the absolute value sign; is the natural exponential function, is the maximum value function. 5.The method of claim 3, wherein, the assignment of the data points in the three-dimensional point cloud data that are not selected as initial clustering clusters to the corresponding basic clustering clusters comprises: calculating the Euclidean distance of any data point that is not selected as an initial clustering cluster to the center of all basic clustering clusters; assigning the data point that is not selected as an initial clustering cluster to the basic clustering cluster closest to it. 6.The method of claim 1, wherein, the evaluation of the damage of the historical site according to the damage scores at all corner points of the plurality of structural regions comprises: when the damage score at the corner point is greater than the preset damage acceptability, marking the corner point; when the proportion of the number of marked corner points to the total number of corner points of the historical site is greater than the preset proportion, marking the historical site as severely damaged. 7.The method of claim 1, wherein, the determination of the first fitting surface and the second fitting surface comprises: determining the mathematical model of the fitting surface, which includes a quadratic surface and a parametric surface; using the least square method to calculate the parameters of the mathematical model to obtain the fitting surface. 8.The method of claim 1, wherein, the collection of three-dimensional point cloud data of the historical site comprises: shooting images of the historical site using aerial photography equipment; constructing the three-dimensional structure of the historical site by MVS algorithm to obtain the three-dimensional point cloud data of the historical site.

Images