Data reconstruction method and system for building house remote sensing surveying and mapping model

Abstract

The invention relates to the technical field of image data processing, in particular to a data reconstruction method and system for a building house remote sensing surveying and mapping model, and the method comprises the steps: obtaining the three-dimensional point cloud data of a building house, and carrying out the initial layering through the local geometric features of the three-dimensional point cloud data, and obtaining all target levels; determining whether false point cloud data exist in the target levels and determining correction parameters of the false point cloud data by utilizing a geometric structure relationship of joint surfaces between the target levels of the adjacent building stages; under the condition that the false point cloud data exists in the target hierarchy, correcting the point cloud data in the target hierarchy by using the correction parameter to obtain corrected point cloud data; and determining a thinning parameter of the target level by using the corrected point cloud data, and determining a target point cloud model of the building house by using the thinning parameter. According to the method, error points or irrelevant points can be effectively removed, the three-dimensional data of the building is ensured to be more suitable for the actual situation, and meanwhile, the processing amount of false point cloud data is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and in particular to a data reconstruction method and system for remote sensing mapping models of buildings. Background Art

[0002] With the rapid development of laser and computer technologies, the use of three-dimensional information from building surfaces for digital city construction is becoming increasingly common. Point cloud technology can provide critical data support for digital city construction, map production, and urban planning, thereby helping people gain a deeper understanding and management of urban space.

[0003] Point cloud data is characterized by high density and precision, but it also presents challenges such as large data volumes and computational complexity. This is particularly true for architectural surveying and mapping, where point cloud data processing and storage efficiency is low. To improve subsequent analysis, model building, and visualization, point cloud data simplification is necessary. Existing point cloud data simplification methods often involve data thinning, but balancing model accuracy and data processing efficiency remains a technical challenge. Excessive data simplification can reduce model accuracy, impacting subsequent architectural analysis and model reconstruction. Summary of the Invention

[0004] In order to solve the technical problem of how to better balance the relationship between building model accuracy and data processing efficiency, the present invention aims to provide a data reconstruction method and system for remote sensing mapping models of buildings. The technical solutions adopted are as follows: The present invention provides a data reconstruction method for a remote sensing mapping model of a building, the method comprising: Obtain 3D point cloud data of the building, and use the local geometric features of the 3D point cloud data to perform initial stratification to obtain each target layer; By using the geometric structure relationship between the joint surfaces of the target layers in adjacent construction stages, it is determined whether there is false point cloud data in the target layer and the correction parameters of the false point cloud data are determined; In the case where false point cloud data exists in the target layer, the point cloud data in the target layer is corrected using the correction parameters to obtain corrected point cloud data; The corrected point cloud data is used to determine the thinning parameters of the target layer, and the thinning parameters are used to determine the target point cloud model of the building.

[0005] Furthermore, the initial stratification using the local geometric features of the three-dimensional point cloud data to obtain each target level includes: Any point cloud data in the three-dimensional point cloud data is used as the central point cloud of the preset spherical area where it is located; Determine a curvature estimation value of the central point cloud, and use the curvature estimation value to determine local geometric features of the central point cloud; The local geometric features and spatial coordinates of the central point cloud are spliced into mixed features, and the mixed features are clustered to obtain each target level through initial stratification.

[0006] Furthermore, determining the curvature estimation value of the central point cloud and using the curvature estimation value to determine the local geometric features of the central point cloud includes: Determine the vector angle between the normal vector of the central point cloud and the normal vectors of other point clouds in the preset spherical area; Using the angles of each vector, the mean of the vector angles is calculated and used as the curvature estimate of the central point cloud; The local geometric features of the central point cloud are calculated using the curvature estimation value of the central point cloud and the curvature estimation values of other point clouds in the preset spherical area.

[0007] Furthermore, the method of using the geometric structure relationship of the joint surface between target levels in adjacent construction stages to determine whether there is false point cloud data in the target level and determining correction parameters for the false point cloud data includes: Determine the joint vectors of the planes where the point cloud data of each target level in the adjacent construction stages are located, and determine the overlap evaluation index of the joint surfaces between the target levels in the adjacent construction stages; The coincidence evaluation index is used to determine whether there is false point cloud data in the target layer, and the correction parameters of the false point cloud data are calculated using the coincidence evaluation index and the joint vector of the plane where the respective point cloud data are located.

[0008] Furthermore, the determination of the overlap evaluation index of the joint surfaces between target levels of adjacent construction stages includes: Using the joint vectors of the planes where the point cloud data of the respective target levels in the adjacent building stages are located, the vector angle between the respective joint vectors is determined; The vector angle is used to determine the geometric relationship similarity between the respective planes in the target level of adjacent building stages; The geometric proximity of the joint surfaces between the target levels of adjacent building stages is determined using the Euclidean distance between the respective reference points in the target levels of adjacent building stages; The geometric relationship similarity and geometric proximity are used to calculate the overlap evaluation index of the joint surfaces between the target levels of adjacent construction stages.

[0009] Furthermore, the method of utilizing the geometric structure relationship of the joint surfaces between target levels in adjacent construction stages to determine whether there is false point cloud data in the target level and determining correction parameters for the false point cloud data may also include: Determine the target level thinned benchmark point set and each benchmark point therein; Determine the outlier degree of the benchmark point, and use the outlier degree to preliminarily screen the benchmark point to obtain the target level after the preliminarily screening update.

[0010] Furthermore, the determination of the target level thinned reference point set and each reference point therein includes: Randomly determine a point in the target layer as the initial reference point, determine the distance to be confirmed between the initial reference point and the current candidate reference point in the target layer, and the average distance between the current candidate reference point and other unselected points; A weighted calculation is performed on the distribution uniformity corresponding to the distance to be confirmed and the average distance to determine the thinned reference point set of the target level and each reference point therein.

[0011] Furthermore, the step of determining the outlier degree of the reference point and using the outlier degree to preliminarily screen the reference point to obtain a target level after the preliminarily screening update includes: Determine the spatial location distance between the reference point and other point clouds in the target layer and the number of point clouds in the target layer; Determine the reference laser reflection intensity of the reference point and other laser reflection intensities of other point clouds; The reflection intensity characteristic factor of the reference point is calculated using the spatial position distance, the number of point clouds, the reference laser reflection intensity and the other laser reflection intensities; The outlier degree of the reference point is calculated using the reflection intensity characteristic factor of the reference point; The outlier degree is compared with the preset outlier threshold, and the benchmark points greater than or equal to the preset outlier threshold are initially screened and removed to obtain the target level after the initial screening update.

[0012] Furthermore, determining thinning parameters of a target layer using the corrected point cloud data, and determining a target point cloud model of a building using the thinning parameters, includes: determining a local geometric feature difference between the corrected point cloud in the target level and the current number of point clouds in the target level; Using the local geometric feature differences and the current number of point clouds, the thinning parameters of the target level are calculated; The target level point cloud is thinned and smoothed using the thinning parameters to obtain the target point cloud model of the building.

[0013] The present invention provides a data reconstruction system for a remote sensing mapping model of a building, for implementing the data reconstruction method for a remote sensing mapping model of a building as described in any one of the above items; the system comprises: The point cloud layering module is used to obtain the 3D point cloud data of the building and house, and use the local geometric features of the 3D point cloud data to perform initial layering to obtain each target layer; A point cloud correction module is used to determine whether there is false point cloud data in the target layer and determine correction parameters for the false point cloud data by using the geometric structure relationship between the joint surfaces of the target layers in adjacent construction stages; if there is false point cloud data in the target layer, the correction parameters are used to correct the point cloud data in the target layer to obtain corrected point cloud data; The thinning calculation module is used to use the corrected point cloud data to determine the thinning parameters of the target layer, and use the thinning parameters to determine the target point cloud model of the building.

[0014] The present invention has the following beneficial effects: In the process of simplifying the existing three-dimensional point cloud data of buildings by layering and thinning, the reflection and refraction of laser light by building windows, glass exterior walls, etc. makes the originally obtained measurement points inaccurate, resulting in a large amount of false point cloud data, which interferes with the point cloud layering results.

[0015] From the perspective of removing false point cloud data, the present invention performs an initial hierarchical division based on the original three-dimensional point cloud data, and sets thinning parameters for each level. By comparing the degree of deviation between the layered thinning results and the building structure characteristics represented by the original point cloud data, it is determined whether there is interference data (false point cloud data) in the current level, and then the false point cloud data is corrected. Furthermore, the present invention corrects the reference points in the point cloud layered thinning process to reduce false point cloud data generated by glass reflections, etc., and corrects the deviation between the layered thinning results and the original building structure. It can effectively remove error points or irrelevant points, ensuring that the three-dimensional data of the building is more in line with the actual situation, while reducing the processing amount of false point cloud data, thereby better balancing the relationship between the model accuracy of the building and the data processing efficiency, and greatly improving the model accuracy of the building and the point cloud data processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 A flowchart of the steps of a data reconstruction method for a remote sensing mapping model of a building provided by one embodiment of the present invention; Figure 2 A detailed flow chart of step S1 in a data reconstruction method for a remote sensing mapping model of a building provided by one embodiment of the present invention; Figure 3A detailed flow chart of step S2 in a data reconstruction method for a remote sensing mapping model of a building provided by one embodiment of the present invention; Figure 4 A detailed flow chart of step S4 in a data reconstruction method for a remote sensing mapping model of a building provided by one embodiment of the present invention; Figure 5 A schematic diagram of the structure of the hardware operating environment of a data reconstruction device for a remote sensing mapping model of a building involved in an embodiment of the present invention; Figure 6 This is a schematic diagram of the framework structure of a data reconstruction system for remote sensing mapping models of buildings involved in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effectiveness of a data reconstruction method for remote sensing mapping models of buildings proposed in accordance with the present invention. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0020] The following describes in detail a specific solution of a data reconstruction method for a remote sensing mapping model of a building provided by the present invention in conjunction with the accompanying drawings.

[0021] Example 1: For the data reconstruction method of the building remote sensing mapping model provided by the present invention, please refer to Figure 1 , which shows a flow chart of the steps of a data reconstruction method for a remote sensing mapping model of a building provided by an embodiment of the present invention.

[0022] The method comprises: Step S1, obtaining three-dimensional point cloud data of a building, and performing initial stratification using local geometric features of the three-dimensional point cloud data to obtain each target layer; First, the specific scenario targeted by this embodiment may be: Mapping buildings at different stages of construction can reveal any positioning or dimensional issues. However, during this mapping process, the presence of reflective surfaces like glass exterior walls and windows often leads to abnormal (3D) point cloud data density or inaccurate positioning at certain locations. This can create interfering false point clouds (referred to as "interference point clouds" or "interference data"), leading to inaccurate assessments of construction deviations. Therefore, it's necessary to monitor point cloud data from different stages of construction, correct for the effects of reflective surfaces on 3D model construction, and obtain accurate remote sensing mapping data.

[0023] In this example, drones and 3D laser scanners are used to acquire 3D point cloud data and laser intensity information for each stage of the building construction process. The construction process (construction stages) includes foundation, structural framework, floor construction, roof and facade construction, and finishing and detailing. The point cloud data includes its position in 3D space, represented by a 3D rectangular coordinate system. Laser intensity information refers to the degree of reflection of the laser beam on the building surface.

[0024] After the collected point cloud data are merged and spliced, there will be some redundant point clouds. In order to ensure the quality of subsequent modeling, the point cloud data needs to be denoised and processed. The two sets of point cloud data in different coordinate systems are aligned, and geometric filtering is used to remove high-deviation points that are completely inconsistent with the building structure.

[0025] The point cloud data obtained at different construction stages are registered using the ICP (Iterative Closest Point) algorithm so that they can be analyzed in a unified coordinate system.

[0026] A large amount of point cloud data is collected at different stages of building construction. Excessive point cloud counts can place additional pressure on subsequent data analysis and model building, reducing model efficiency and the resulting processing results. Therefore, in practical applications, it is necessary to thin out and simplify the point cloud data to reduce the corresponding number of points. However, due to interference from reflective surfaces such as glass exterior walls and windows, the collected raw point cloud data may contain some false point cloud data, also known as interference point cloud data or noise data, which can affect the effectiveness of data thinning and simplification. Therefore, this embodiment performs an initial hierarchical division based on the local geometric features of the raw point cloud data (3D point cloud data) at each stage to obtain target levels. By analyzing the matching results and fusion levels of the same level at different stages, the degree of deviation between the building structural characteristics represented by the hierarchical thinning results and the original point cloud data (specifically, the geometric structural relationship between the joints between target levels of adjacent construction stages) is compared to determine whether there is interference data in the target level of the current construction stage (hereinafter referred to as the "current level"), thereby correcting the false point cloud data.

[0027] Specifically, in one embodiment, please refer to Figure 2 , the step S1 comprises: Step S11, taking any point cloud data in the three-dimensional point cloud data as the center point cloud of the preset spherical area in which it is located; Step S12, determining a curvature estimation value of the center point cloud, and using the curvature estimation value to determine local geometric features of the center point cloud; The step S12 specifically includes: Determine the vector angle between the normal vector of the central point cloud and the normal vectors of other point clouds in the preset spherical area; Using the angles of each vector, the mean of the vector angles is calculated and used as the curvature estimate of the central point cloud; The local geometric features of the central point cloud are calculated using the curvature estimation value of the central point cloud and the curvature estimation values of other point clouds in the preset spherical area.

[0028] Buildings have various architectural features, such as eaves, windows, relief shapes and other characteristic details. In the three-dimensional point cloud data acquired at different times, the density and expression form of the point cloud data are different at different locations of the building. If the same thinning parameters are set for the entire building, the characteristic details cannot be fully and effectively preserved. Therefore, to ensure that the characteristic point clouds or details are not affected during the point cloud simplification process, this embodiment combines the differences in the expression characteristics of point cloud data at different locations and first performs layered processing on the point cloud data before simplification.

[0029] Set the spherical sliding window with radius r=3 as the preset spherical area, and for each point cloud data placed at the center of the spherical area , calculate the center point (cloud) Normal vector With every other neighboring point in the spherical neighborhood Normal vector The vector angle between .

[0030] When a point cloud in a building is in a corner or detail area, due to the large degree of surface shape change, the difference between the angles in the neighborhood where the point cloud is located is more obvious. Therefore, for several vector angles in the current spherical neighborhood, , calculate the mean of the vector angles of the point cloud data in the spherical neighborhood as the curvature estimate of the central point cloud data: , traverse to obtain the curvature estimation value of all point cloud data, Indicates the center point Normal vector With the center point The vector angle between the normal vectors of other j-th point cloud data in the spherical area, Represents central point cloud data The estimated curvature of .

[0031] Since the curvature estimates of the center points may be similar when located in different areas of the building components, such as flat walls, curved roofs, edge beams and columns, this step uses the statistics of the point cloud curvature in the spherical neighborhood where the current center point cloud is located to comprehensively construct the center point. Local geometric features of: , Indicates the center point The curvature estimation value of other j-th point cloud data in the spherical area, Center Point The number of other point clouds in the spherical area, Represents central point cloud data The local geometric features of the point cloud within the neighborhood are larger when the curvature estimates of the point cloud data within the neighborhood vary. The local geometric features of the point cloud surface within a spherical neighborhood reflect the degree of shape change of the surface surrounding the point. When the degree of change of the surrounding surface is large, the local geometric features will be correspondingly larger.

[0032] In step S13, the local geometric features and spatial coordinates of the central point cloud are spliced into mixed features, and the mixed features are clustered to obtain each target level by initial stratification.

[0033] According to each center point , its local geometric features Its spatial coordinates Splicing into mixed features: , the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) density clustering algorithm is used for cluster analysis, and the different levels generated by the clustering results represent the target levels of different building components in the building house.

[0034] Step S2, using the geometric structure relationship of the joint surface between the target layers of adjacent construction stages, determining whether there is false point cloud data in the target layer and determining correction parameters for the false point cloud data; By comparing the differences between the newly added structures and the original structures during the construction process, mainly referring to the geometric structure differences of the joint surfaces after stratification, it is possible to analyze whether there are problems with position or size deviations in the building during the construction process. The traditional method of retaining reference points in a hierarchy is to select the center point or the farthest point of the spatial position in the same hierarchy as the reference point. These reference points can effectively represent the geometric shape of the data, but due to the presence of reflective surfaces such as windows and glass exterior walls, the reflection and refraction of the glass surface will produce some false point clouds. This selection method does not take into account that some data points are false point clouds, which makes the extracted hierarchical characteristics have errors. Therefore, this embodiment further refines the initial stratification results. After completing the stratification of the point cloud data, it is necessary to combine the point cloud data of each layer obtained in different stages for analysis. By calculating the geometric difference between the stratified point cloud data and the original point cloud data, it is determined whether there are false point clouds in the current hierarchy, and accurate stratification reference points are obtained by correcting the false point clouds.

[0035] In one embodiment, to facilitate understanding of the specific process of step S2, before step S2, the method further includes: Determine the target level thinned benchmark point set and each benchmark point therein; Determine the outlier degree of the benchmark point, and use the outlier degree to preliminarily screen the benchmark point to obtain the target level after the preliminarily screening update.

[0036] In one embodiment, specifically, determining the thinned reference point set of the target level and each reference point therein includes: Randomly determine a point in the target layer as the initial reference point, determine the distance to be confirmed between the initial reference point and the current candidate reference point in the target layer, and the average distance between the current candidate reference point and other unselected points; A weighted calculation is performed on the distribution uniformity corresponding to the distance to be confirmed and the average distance to determine the thinned reference point set of the target level and each reference point therein.

[0037] Since the local geometric information of the point cloud corresponding to different construction steps is not exactly the same during the construction process, the stratification results of different construction stages are different when the point cloud data is stratified. Therefore, we first perform a differential based on the point cloud data of two adjacent construction stages. The point cloud data corresponding to the difference is the newly built or changed part of the building in that construction stage. The newly built part structure in the construction process is extracted, and the initial point cloud stratification results are further subdivided according to the newly built part. For example, if there is a new structure in a certain initial "wall" stratification, the original "wall" level needs to be further subdivided into two sub-levels, "wall" and "new structure". That is, the target level can be subdivided into sub-levels, or the target level can be considered to include some sub-levels.

[0038] Extract the reference points of each sub-level separately. Randomly select a point in the V-th sub-level point cloud data as the initial reference point, and then select the point farthest from the selected point set as the next reference point in turn, until a point that repeats the selected reference point appears and the iteration stops; at the same time, calculate the distance between each point cloud and other points in the level, and select the point with the most uniform spatial distribution as the reference point to avoid excessive thinning of dense areas. This process is relatively easy to understand. Randomly select a point in the V-th sub-level point cloud data as the initial reference point, calculate the distance to be confirmed between each point in the level (the current candidate reference point) and the initial reference point, and the average distance between the current candidate reference point and other unselected points in the target level, and select a suitable point as the next reference point according to the rule of "the farthest distance in the distance to be confirmed + the most uniform distribution corresponding to the average distance" (ideally, the point with the farthest distance and the most uniform distribution is the best reference point) until a point that repeats the selected reference point appears and the iteration stops. Since there are generally few benchmark points that meet the "longest distance + most even distribution" criteria, we can assign a weight to each of these two criteria based on the maximum distance. This way, the longer the distance to be confirmed, the smaller the weight for distribution uniformity, and the larger the weight for the distance to be confirmed. Conversely, the closer the distance to be confirmed, the larger the weight for distribution uniformity, and the smaller the weight for the distance to be confirmed. The distribution uniformity of a point is determined by calculating the average distance between the current candidate benchmark point and the other unselected points in the target layer.

[0039] Arrange the obtained reference points to obtain the sparse reference point set under the sub-layer of the current level , where n is the number of all reference points obtained in the Vth level. Then, the thinned reference point set in each target level and each reference point therein are determined.

[0040] In one embodiment, specifically, determining the outlier degree of the reference point and using the outlier degree to preliminarily screen the reference point to obtain a preliminarily screened and updated target level includes: Determine the spatial location distance between the reference point and other point clouds in the target layer and the number of point clouds in the target layer; Determine the reference laser reflection intensity of the reference point and other laser reflection intensities of other point clouds; The reflection intensity characteristic factor of the reference point is calculated using the spatial position distance, the number of point clouds, the reference laser reflection intensity and the other laser reflection intensities; The outlier degree of the reference point is calculated using the reflection intensity characteristic factor of the reference point; The outlier degree is compared with the preset outlier threshold, and the benchmark points greater than or equal to the preset outlier threshold are initially screened and removed to obtain the target level after the initial screening update.

[0041] Since some false point cloud data are usually isolated in the hierarchy and show a significantly larger distance from their neighboring points, they will be screened out when screening benchmark points because of their longest distance. As a result, false point clouds exist in the thinning results that retain the benchmark points, making it impossible to obtain accurate building characteristics when performing feature analysis based on the thinning results.

[0042] The reflection intensity of the laser varies depending on the building surface material (such as metal, concrete, etc.). Therefore, it is necessary to calculate the reflection intensity characteristic factor of the laser reflection intensity corresponding to the reference point at the same level: in, Indicates the The reflection intensity characteristic factor of each reference point, Indicates the number of points contained in the Vth level (number of point clouds); Indicates that other points and the current reference point The spatial position distance between them is the Euclidean distance formula, It can be considered as the similarity calculation of spatial position distance. The larger the distance, the smaller the similarity. It reflects the similarity between point p and the current reference point. The greater the spatial distance between the points, the closer the point p is to the current reference point. The smaller the impact; Indicates the The maximum value of the laser reflection intensity at multiple angles corresponding to the reference points; Indicates other Point and The benchmark points corresponds to the laser reflection intensity at the same angle.

[0043] If the reflection intensity characteristic factor of the current reference point is the same or similar to the corresponding reflection intensity characteristic factors of multiple points in the same level at the same angle, then the current reference point is more likely to be in the same plane as the other points, indicating that the current reference point is more likely to belong to normal point cloud data. Therefore, the outlier degree of the reference point in the same level is calculated: ,in, Indicates the The outlier degree of the reference point, Indicates the The reflection intensity characteristic factor of each reference point; Indicates the reference point The mean value of the reflection intensity characteristic factor of all reference points in the layer; Indicates the number of benchmark points screened within level V; Indicates the The first one filtered out from the level where the benchmark points are located The reflection intensity characteristic factor of the reference point. In the outlier degree formula, a hyperparameter can be set as needed to prevent the denominator from being zero, such as 0.00001.

[0044] According to the calculated outlier degree of the benchmark points in the hierarchy, a preset outlier threshold T=0.75 (which can be adjusted according to actual needs) is set to thin out the benchmark point set in the same hierarchy. Perform screening and remove the benchmark points with outliers greater than or equal to 0.75 to obtain the target level after preliminary screening and update.

[0045] Specifically, in one embodiment, please refer to Figure 3 , adjacent construction stages can be divided into the previous construction stage and the current construction stage; the step S2 includes: Step S21, determining the joint vectors of the planes where the point cloud data of the respective target levels of the adjacent construction stages are located, and determining the overlap evaluation index of the joint surfaces between the target levels of the adjacent construction stages; The step S21 specifically includes: Using the joint vectors of the planes where the point cloud data of the respective target levels in the adjacent building stages are located, the vector angle between the respective joint vectors is determined; The vector angle is used to determine the geometric relationship similarity between the respective planes in the target level of adjacent building stages; The geometric proximity of the joint surfaces between the target levels of adjacent building stages is determined using the Euclidean distance between the respective reference points in the target levels of adjacent building stages; The geometric relationship similarity and geometric proximity are used to calculate the overlap evaluation index of the joint surfaces between the target levels of adjacent construction stages.

[0046] After deleting the false point cloud data in the benchmark points, some false point clouds still exist in the hierarchy. Therefore, it is necessary to generate an updated reference model based on the point cloud data obtained in each construction stage.

[0047] Due to the different current construction stages, the corresponding stratification results are different. In the multi-level point cloud data, different splicing surfaces or components will have different structural characteristics, resulting in different connection methods of the building point cloud data. The position and connection direction of the splicing surface between the reference points of these newly added sub-layers and the original construction structure of the previous stage are calculated to correct the false point cloud data in the currently obtained newly added sub-layers.

[0048] According to the coordinates of the reference points retained in the three-dimensional coordinate system, the cross product formula is used to calculate the normal vectors of several reference points in the sublayer located in the same plane as the joint vector of the plane , V represents level V.

[0049] For each sub-layer of the building stage, calculate the joint vector of the current plane (corresponding to the current building stage) of the sub-layer of the same plane in the current level The junction vector with any plane of the original level (corresponding to the previous building stage) The vector angle between , using cosine similarity to obtain multiple angle similarities between each joining vector in the current plane and the joining vector of the original level, as the geometric relationship similarity between the current plane in the sub-layer and the original level: in, Indicates the geometric similarity, represents the vector angle, Indicates the number of sub-levels V.

[0050] The cosine value of the angle between the combined vector of the plane formed by several reference points in the current sub-layer and the original level plane The closer it is to 1, the better the connection effect between layers in the collected point cloud data is, and the smaller the possibility of false point cloud data existing in the layer. L represents the layer L.

[0051] According to the current sub-layer's joint vector Join vector with the original level The angle between , defines the distance between several reference points between two levels. When the cosine value of the angle is closer to 1, it means that the possibility of the two reference points being parallel is greater. At this time, the Euclidean distance between the two reference points is calculated and recorded as , calculate the mean Euclidean distance between each pair of planes in the two levels, and select the minimum distance as the geometric proximity of the interface between the two levels: ,By calculating the minimum Euclidean distance of the joint surface between two levels, the geometric proximity between,the joint surfaces is determined.

[0052] Theoretically, the interface between the newly built building components and the original layer should completely overlap. If the interface between the current sub-layer and the original layer partially overlaps, it means that there may be false point cloud data in the sub-layer. When the geometric connection relationship value is closer to 1, the mean distance between the two layers is smaller, indicating that the overlap between the layers is higher. The overlap evaluation index of the interface between the two layers is obtained: in, represents the overlap evaluation index, Represents the similarity of geometric relationships between levels; Indicates the minimum distance between the planes of the two levels, that is, the geometric proximity. The smaller the distance between the joint surfaces of the two levels, The closer the value is to 1, the higher the geometric consistency of the data between the two levels, which means the higher the overlap between the two joint surfaces. The lower the value, the poorer the overlap of the data between the two levels, and the higher the possibility of false point cloud data.

[0053] In step S22 , the coincidence evaluation index is used to determine whether there is false point cloud data in the target layer, and correction parameters of the false point cloud data are calculated using the coincidence evaluation index and the joint vectors of the planes where the respective point cloud data are located.

[0054] Here we can also set the overlap threshold to 0.75. Normalization is performed, and the normalized coincidence evaluation index and coincidence threshold are compared. For the level less than 0.75, it is considered that there is false point cloud data, and then the false point cloud of the level is corrected.

[0055] The overlap relationship of the joint surfaces between the layers is verified based on the overlap evaluation index. When the joint surfaces between the newly added components and the original structure do not completely overlap, it means that there are false point clouds in the current newly added structure that affect the joint surface relationship. The position of the false point clouds needs to be readjusted to make the joint surface relationship more consistent with the actual situation. Therefore, the correction parameters of the false point clouds are obtained: ,in represents the correction parameter, represents the overlap evaluation index, Indicates the joining vector of the plane where the i-th point cloud data in the sublayer of the current level is located. Represents the joining vector of the plane where the L-th point cloud data in the sub-layer of the previous building stage is located, Indicates the magnitude of the difference vector.

[0056] Step S3, when there is false point cloud data in the target layer, correcting the point cloud data in the target layer using the correction parameters to obtain corrected point cloud data; According to the correction parameters of the benchmark points obtained by the merchant, if the influence coefficients of multiple adjacent different point cloud levels are large, the corresponding retention degree of this level is higher, that is, it should be given a higher weight coefficient in the subsequent benchmark point adjustment process. This setting will allow the levels with similar influence to have a higher correction weight for the current false point cloud, thereby achieving the purpose of correcting the false point cloud: .in is the three-dimensional coordinate of the original false point cloud, is the three-dimensional coordinate of the corrected point cloud, that is, the corrected point cloud data, Indicates the correction parameter.

[0057] After correcting the false point cloud data, the initial stratification results of the building are re-optimized and stratified. After each correction, the geometric deviation of the point cloud data is recalculated and analyzed to ensure that the interfering data is fully corrected and the characteristics of each level of the building can be accurately presented.

[0058] Step S4: using the corrected point cloud data to determine the thinning parameters of the target layer, and using the thinning parameters to determine the target point cloud model of the building.

[0059] For details, please refer to Figure 4 , the step S4 comprises: Step S41, determining the local geometric feature differences between the corrected point clouds in the target level and the current number of point clouds in the target level; Step S42, using the local geometric feature differences and the current number of point clouds, calculate the thinning parameters of the target level; In step S43 , the target level is thinned and smoothed using the suction parameters to obtain a target point cloud model of the building.

[0060] According to the variation range of the local geometric features of each layer of point cloud data after correction, different thinning parameters are set for different levels: Point cloud data at different levels represent different building components in a building. For areas with small changes in surface details, such as smooth walls and evenly varying roofs, a uniform thinning method can be used to remove point cloud data from a large area while effectively reducing data redundancy. For areas with large changes in surface details, such as corners, windows, and glass exterior wall shapes, since the areas contain more detailed information, it may be necessary to retain a large amount of the level and use a non-uniform thinning method with a set threshold. Calculate the thinning parameters for the Vth level ,in represents the thinning parameter, Represents the local geometric feature differences between point cloud data in the hierarchy, where Represents the local geometric features of the i-th point in the V-th level; represents the local geometric features of the i+1th point in the Vth level, nc represents the number of points contained in the current level (the current number of point clouds), and the denominator is increased by one to prevent the difference in local geometric features from being zero and affecting the calculation results.

[0061] Calculate the thinning parameters between each level, perform point cloud thinning on each level, and obtain the thinning results of the level.

[0062] After completing the thinning of each layer of point cloud data, the point cloud data at different levels are color-coded separately. For areas with obvious cracks or jagged shapes, such as the connection between the wall and the roof, the point cloud is smoothed using the moving least squares method. The simplification effect of the building model is judged by calculating the ratio between the number of retained points after point cloud simplification and the number of original points. When the ratio is less than or equal to 0.3 and greater than or equal to 0.1, it means that the current simplification effect is good, and the final simplified point cloud model is generated.

[0063] The present invention starts from the perspective of removing false point cloud data. The present invention performs an initial hierarchical division based on the original three-dimensional point cloud data, and sets thinning parameters for each level. By comparing the degree of deviation between the layered thinning results and the building structure characteristics represented by the original point cloud data, it determines whether there is interference data (false point cloud data) in the current level, and then corrects the false point cloud data. Furthermore, the present invention corrects the reference points in the point cloud layered thinning process to reduce false point cloud data generated by glass reflections, etc., and corrects the deviation between the layered thinning results and the original building structure. It can effectively remove error points or irrelevant points, ensuring that the three-dimensional data of the building is more in line with the actual situation, while reducing the processing amount of false point cloud data, thereby better balancing the relationship between the model accuracy of the building and the data processing efficiency, so that the model accuracy of the building and the point cloud data processing efficiency are greatly improved.

[0064] Example 2: The embodiment of the present invention further provides a data reconstruction device for remote sensing mapping model of a building. The data reconstruction device for remote sensing mapping model of a building can be a data computing and processing device such as a computer, a server, or a combination of multiple devices.

[0065] like Figure 5 As shown, Figure 5 It is a structural diagram of the hardware operating environment of a data reconstruction device for a remote sensing mapping model of a building involved in an embodiment of the present invention.

[0066] like Figure 5As shown, the data reconstruction device for remote sensing mapping models of buildings and houses may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Communication bus 1002 is used to enable communication between these components. User interface 1003 may include a display and an input unit, such as a control panel. Optionally, user interface 1003 may also include a standard wired interface or a wireless interface. Network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). Memory 1005 may be high-speed RAM or non-volatile memory, such as a disk drive. Memory 1005 may also be a storage device independent of processor 1001. Memory 1005, a computer storage medium, may include a data reconstruction program for remote sensing mapping models of buildings and houses.

[0067] Those skilled in the art will understand that Figure 5 The hardware structure shown in the figure does not constitute a limitation of the device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0068] Continue to refer to Figure 5 , Figure 5 The memory 1005 as a computer-readable storage medium may include an operating system, a user interface module, a network communication module, and a data reconstruction program for the remote sensing mapping model of a building.

[0069] exist Figure 5 In the embodiment, the network communication module is mainly used to connect to the server and can communicate data with the server; and the processor 1001 can call the data reconstruction program of the remote sensing mapping model of the building house stored in the memory 1005 and execute the steps in the above embodiments.

[0070] The hardware structure of the data reconstruction device based on the above-mentioned remote sensing mapping model of a building is used to implement various embodiments of the data reconstruction method of the remote sensing mapping model of a building in the present invention.

[0071] In addition, the present invention also provides a data reconstruction system for remote sensing mapping models of buildings, please refer to Figure 6 The data reconstruction system of the building remote sensing mapping model includes: Point cloud layering module A10 is used to obtain 3D point cloud data of buildings and houses, and perform initial layering using the local geometric features of the 3D point cloud data to obtain each target layer; The point cloud correction module A20 is configured to use the geometric structural relationship between the joint surfaces of the target layers in adjacent construction stages to determine whether there is false point cloud data in the target layer and determine correction parameters for the false point cloud data; if there is false point cloud data in the target layer, the point cloud data in the target layer is corrected using the correction parameters to obtain corrected point cloud data; The thinning calculation module A30 is used to determine the thinning parameters of the target layer using the corrected point cloud data, and determine the target point cloud model of the building using the thinning parameters.

[0072] Furthermore, the point cloud layering module A10 is further configured to: Any point cloud data in the three-dimensional point cloud data is used as the central point cloud of the preset spherical area where it is located; Determine a curvature estimation value of the central point cloud, and use the curvature estimation value to determine local geometric features of the central point cloud; The local geometric features and spatial coordinates of the central point cloud are spliced into mixed features, and the mixed features are clustered to obtain each target level through initial stratification.

[0073] Furthermore, the point cloud layering module A10 is further configured to: Determine the vector angle between the normal vector of the central point cloud and the normal vectors of other point clouds in the preset spherical area; Using the angles of each vector, the mean of the vector angles is calculated and used as the curvature estimate of the central point cloud; The local geometric features of the central point cloud are calculated using the curvature estimation value of the central point cloud and the curvature estimation values of other point clouds in the preset spherical area.

[0074] Furthermore, the point cloud correction module A20 is further configured to: Determine the joint vectors of the planes where the point cloud data of each target level in the adjacent construction stages are located, and determine the overlap evaluation index of the joint surfaces between the target levels in the adjacent construction stages; The coincidence evaluation index is used to determine whether there is false point cloud data in the target layer, and the correction parameters of the false point cloud data are calculated using the coincidence evaluation index and the joint vector of the plane where the respective point cloud data are located.

[0075] Furthermore, the point cloud correction module A20 is further configured to: Using the joint vectors of the planes where the point cloud data of the respective target levels in the adjacent building stages are located, the vector angle between the respective joint vectors is determined; The vector angle is used to determine the geometric relationship similarity between the respective planes in the target level of adjacent building stages; The geometric proximity of the joint surfaces between the target levels of adjacent building stages is determined using the Euclidean distance between the respective reference points in the target levels of adjacent building stages; The geometric relationship similarity and geometric proximity are used to calculate the overlap evaluation index of the joint surfaces between the target levels of adjacent construction stages.

[0076] Furthermore, the point cloud correction module A20 is further configured to: Determine the target level thinned benchmark point set and each benchmark point therein; Determine the outlier degree of the benchmark point, and use the outlier degree to preliminarily screen the benchmark point to obtain the target level after the preliminarily screening update.

[0077] Furthermore, the point cloud correction module A20 is further configured to: Randomly determine a point in the target layer as the initial reference point, determine the distance to be confirmed between the initial reference point and the current candidate reference point in the target layer, and the average distance between the current candidate reference point and other unselected points; A weighted calculation is performed on the distribution uniformity corresponding to the distance to be confirmed and the average distance to determine the thinned reference point set of the target level and each reference point therein.

[0078] Furthermore, the point cloud correction module A20 is further configured to: Determine the spatial location distance between the reference point and other point clouds in the target layer and the number of point clouds in the target layer; Determine the reference laser reflection intensity of the reference point and other laser reflection intensities of other point clouds; The reflection intensity characteristic factor of the reference point is calculated using the spatial position distance, the number of point clouds, the reference laser reflection intensity and the other laser reflection intensities; The outlier degree of the reference point is calculated using the reflection intensity characteristic factor of the reference point; The outlier degree is compared with the preset outlier threshold, and the benchmark points greater than or equal to the preset outlier threshold are initially screened and removed to obtain the target level after the initial screening update.

[0079] Furthermore, the thinning calculation module A30 is further configured to: determining a local geometric feature difference between the corrected point cloud in the target level and the current number of point clouds in the target level; Using the local geometric feature differences and the current number of point clouds, the thinning parameters of the target level are calculated; The target level point cloud is thinned and smoothed using the thinning parameters to obtain the target point cloud model of the building.

[0080] The specific implementation of the data reconstruction system for remote sensing mapping models of buildings and houses of the present invention is substantially the same as the embodiments of the data reconstruction method for remote sensing mapping models of buildings and houses described above, and will not be described in detail here.

[0081] The present invention also provides a computer-readable storage medium. The computer-readable storage medium stores a data reconstruction program for remote sensing mapping models of buildings and houses. When executed by a processor, the data reconstruction program for remote sensing mapping models of buildings and houses implements the steps of the aforementioned data reconstruction method for remote sensing mapping models of buildings and houses.

[0082] The method implemented when the data reconstruction program for the remote sensing mapping model of a building is executed can refer to the various embodiments of the data reconstruction method for the remote sensing mapping model of a building of the present invention, and will not be described in detail here.

[0083] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0084] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from the reference embodiment.

[0085] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] The above description is only a preferred embodiment of the present invention and does not limit the scope of protection of the present invention. All equivalent structural / method transformations made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or direct / indirect application in reference to related technical fields are included in the scope of protection of the present invention.

Claims

1. A data reconstruction method for remote sensing mapping model of a building, characterized in that: The method comprises: Obtain 3D point cloud data of the building, and use the local geometric features of the 3D point cloud data to perform initial stratification to obtain each target level; By using the geometric structure relationship between the joint surfaces of the target layers in adjacent construction stages, it is determined whether there is false point cloud data in the target layer and the correction parameters of the false point cloud data are determined; In the case where false point cloud data exists in the target layer, the point cloud data in the target layer is corrected using the correction parameters to obtain corrected point cloud data; The corrected point cloud data is used to determine the thinning parameters of the target layer, and the thinning parameters are used to determine the target point cloud model of the building.

2. The data reconstruction method of the building remote sensing mapping model according to claim 1 is characterized in that: The initial stratification using the local geometric features of the three-dimensional point cloud data to obtain each target level includes: Any point cloud data in the three-dimensional point cloud data is used as the central point cloud of the preset spherical area where it is located; Determine a curvature estimation value of the central point cloud, and use the curvature estimation value to determine local geometric features of the central point cloud; The local geometric features and spatial coordinates of the central point cloud are spliced into mixed features, and the mixed features are clustered to obtain each target level through initial stratification.

3. The data reconstruction method of the building remote sensing mapping model according to claim 2 is characterized in that: Determining the curvature estimation value of the central point cloud and determining the local geometric features of the central point cloud using the curvature estimation value includes: Determine the vector angle between the normal vector of the central point cloud and the normal vectors of other point clouds in the preset spherical area; Using the angles of each vector, the mean of the vector angles is calculated and used as the curvature estimate of the central point cloud; The local geometric features of the central point cloud are calculated using the curvature estimation value of the central point cloud and the curvature estimation values of other point clouds in the preset spherical area.

4. The data reconstruction method of the building remote sensing mapping model according to claim 1 is characterized in that: The method of utilizing the geometric structure relationship of the joint surfaces between target levels in adjacent construction stages to determine whether there is false point cloud data in the target level and determining correction parameters for the false point cloud data includes: Determine the joint vectors of the planes where the point cloud data of each target level in the adjacent construction stages are located, and determine the overlap evaluation index of the joint surfaces between the target levels in the adjacent construction stages; The coincidence evaluation index is used to determine whether there is false point cloud data in the target layer, and the correction parameters of the false point cloud data are calculated using the coincidence evaluation index and the joint vector of the plane where the respective point cloud data are located.

5. The data reconstruction method of the building remote sensing mapping model according to claim 4 is characterized in that: The determination of the overlap evaluation index of the joint surfaces between target levels of adjacent construction stages includes: Using the joint vectors of the planes where the point cloud data of the respective target levels in the adjacent building stages are located, the vector angle between the respective joint vectors is determined; The vector angle is used to determine the geometric relationship similarity between the respective planes in the target level of adjacent building stages; The geometric proximity of the joint surfaces between the target levels of adjacent building stages is determined using the Euclidean distance between the respective reference points in the target levels of adjacent building stages; The geometric relationship similarity and geometric proximity are used to calculate the overlap evaluation index of the joint surfaces between the target levels of adjacent construction stages.

6. The data reconstruction method of the building remote sensing mapping model according to claim 1 is characterized in that: The method of utilizing the geometric structure relationship of the joint surfaces between target levels of adjacent construction stages to determine whether there is false point cloud data in the target level and determining correction parameters for the false point cloud data may also include: Determine the target level thinned benchmark point set and each benchmark point therein; Determine the outlier degree of the benchmark point, and use the outlier degree to preliminarily screen the benchmark point to obtain the target level after the preliminarily screening update.

7. The data reconstruction method of the building remote sensing mapping model according to claim 6 is characterized in that: The determination of the target level thinned reference point set and each reference point therein includes: Randomly determine a point in the target layer as the initial reference point, determine the distance to be confirmed between the initial reference point and the current candidate reference point in the target layer, and the average distance between the current candidate reference point and other unselected points; A weighted calculation is performed on the distribution uniformity corresponding to the distance to be confirmed and the average distance to determine the thinned reference point set of the target level and each reference point therein.

8. The data reconstruction method of the building remote sensing mapping model according to claim 6 is characterized in that: Determining the outlier degree of the reference point and using the outlier degree to preliminarily screen the reference point to obtain a target level after the preliminarily screening update includes: Determine the spatial location distance between the reference point and other point clouds in the target layer and the number of point clouds in the target layer; Determine the reference laser reflection intensity of the reference point and other laser reflection intensities of other point clouds; The reflection intensity characteristic factor of the reference point is calculated using the spatial position distance, the number of point clouds, the reference laser reflection intensity and the other laser reflection intensities; The outlier degree of the reference point is calculated using the reflection intensity characteristic factor of the reference point; The outlier degree is compared with the preset outlier threshold, and the benchmark points greater than or equal to the preset outlier threshold are initially screened and removed to obtain the target level after the initial screening update.

9. The data reconstruction method of a building remote sensing mapping model according to claim 1, characterized in that: The method of determining thinning parameters of a target layer using the corrected point cloud data and determining a target point cloud model of a building using the thinning parameters includes: determining a local geometric feature difference between the corrected point cloud in the target level and the current number of point clouds in the target level; Using the local geometric feature differences and the current number of point clouds, the thinning parameters of the target level are calculated; The target level point cloud is thinned and smoothed using the thinning parameters to obtain the target point cloud model of the building.

10. A data reconstruction system for remote sensing mapping models of buildings, characterized in that: The system is used to implement the data reconstruction method of the building remote sensing mapping model according to any one of claims 1 to 9; the system comprises: The point cloud layering module is used to obtain the 3D point cloud data of the building and house, and use the local geometric features of the 3D point cloud data to perform initial layering to obtain each target layer; A point cloud correction module is used to determine whether there is false point cloud data in the target layer and determine correction parameters for the false point cloud data by using the geometric structure relationship between the joint surfaces of the target layers in adjacent construction stages; if there is false point cloud data in the target layer, the correction parameters are used to correct the point cloud data in the target layer to obtain corrected point cloud data; The thinning calculation module is used to use the corrected point cloud data to determine the thinning parameters of the target layer, and use the thinning parameters to determine the target point cloud model of the building.

Images