Existing building multi-source point cloud fusion method, device, equipment and medium

Abstract

This invention relates to the field of building surveying, and provides a method, apparatus, equipment, and medium for fusing multi-source point clouds of existing buildings. The method includes: scanning the existing building using multiple scanning devices to obtain raw point cloud data; registering the raw point cloud data based on a preset registration strategy to obtain first point cloud data; filtering the first point cloud data according to a preset filtering strategy to obtain second point cloud data; and fusing the second point cloud data based on a preset fusion strategy to obtain target point cloud data. This invention uses multiple scanning devices to obtain raw point cloud data from existing buildings, and then registers, filters, and fuses the raw point cloud data to obtain target point cloud data. This allows for the processing of point clouds with low overlap or significant density differences, while maintaining data integrity in key areas, ensuring both accuracy and completeness of the point cloud data. It also effectively smooths the boundaries between point clouds from different sources, thereby improving measurement efficiency.

Description

Technical Field

[0001] This invention relates to the field of building surveying, and in particular to a method, apparatus, equipment and medium for multi-source point cloud fusion of existing buildings. Background Technology

[0002] As my country's urbanization process shifts from incremental expansion to stock improvement, urban renewal has become a core strategy for promoting sustainable urban development. Old neighborhoods, as important carriers of urban historical context, present both a key focus and a challenge in urban renewal. However, such projects generally face problems such as the age of the buildings, the significant loss of original drawings, or inconsistencies with the current state, making traditional design and construction methods relying on two-dimensional drawings unsustainable. Three-dimensional point cloud technology offers a crucial solution to the surveying challenges in the renovation of old neighborhoods. A point cloud model is an explicit three-dimensional representation that uses dense sets of points carrying attributes such as location and color to directly describe the surface of an object. In the renovation of existing buildings, a complete point cloud model can intuitively display the current state of the building, accurately measure geometric dimensions, and assist in quality assessment and deformation analysis, providing a reliable data foundation for the current state assessment and renovation design of old neighborhoods.

[0003] Currently, the mainstream technologies for acquiring point clouds mainly include 3D laser scanning based on LiDAR and photogrammetry based on multi-view images. 3D laser scanning, as an active measurement technology, has advantages such as extremely high accuracy, strong anti-interference ability, and minimal impact from lighting conditions; however, its limitations include expensive equipment, relatively limited operational efficiency, and generally poor quality of the acquired texture information. Photogrammetry, on the other hand, utilizes Structure of Motion (SfM) and Multi-View Stereo Reconstruction (MVS) algorithms to calculate camera pose and scene depth from multi-view images. This technology is low-cost, flexible, and can acquire high-resolution textures with rich colors and realistic details; however, its accuracy is relatively low, and the reconstruction effect heavily depends on stable lighting conditions and the rich texture features of the object's surface.

[0004] Therefore, the inherent limitations of the aforementioned single technical approach make it difficult to independently address the complex multi-scale and multi-objective requirements of old neighborhood renovation projects, and it is difficult to balance accuracy, integrity, and high-quality texture, resulting in low measurement efficiency. Summary of the Invention

[0005] This invention provides a method, apparatus, equipment, and medium for multi-source point cloud fusion of existing buildings, aiming to solve the problem of low measurement efficiency in existing multi-source point cloud fusion methods for existing buildings.

[0006] To address the aforementioned problems, in a first aspect, embodiments of the present invention provide a method for fusing multi-source point clouds of existing buildings, the method comprising: Raw point cloud data was obtained by scanning existing buildings using various scanning devices; The original point cloud data is registered based on a preset registration strategy to obtain the first point cloud data. The first point cloud data is filtered and processed according to a preset filtering strategy to obtain the second point cloud data; The target point cloud data is obtained by fusing the second point cloud data based on a preset fusion strategy.

[0007] Secondly, embodiments of the present invention also provide a multi-source point cloud fusion device for existing buildings, the multi-source point cloud fusion device for existing buildings comprising: The scanning unit is used to scan existing buildings using various scanning devices to obtain raw point cloud data; A registration unit is used to perform registration processing on the original point cloud data based on a preset registration strategy to obtain the first point cloud data. The filtering unit is used to filter the first point cloud data according to a preset filtering strategy to obtain the second point cloud data. The fusion unit is used to perform fusion processing on the second point cloud data based on a preset fusion strategy to obtain the target point cloud data.

[0008] Thirdly, embodiments of the present invention also provide an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect above.

[0009] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, can implement the method described in the first aspect.

[0010] This invention provides a method, apparatus, device, and medium for fusing multi-source point clouds of existing buildings. The method includes: scanning the existing building using multiple scanning devices to obtain original point cloud data; registering the original point cloud data based on a preset registration strategy to obtain first point cloud data; filtering the first point cloud data according to a preset filtering strategy to obtain second point cloud data; and fusing the second point cloud data based on a preset fusion strategy to obtain target point cloud data. It is understood that this invention utilizes the multiple scanning devices to scan the existing building to obtain the original point cloud data, and then performs registration, filtering, and fusion processing on the original point cloud data to obtain the target point cloud data. This enables the processing of point clouds with low overlap or significant density differences, effectively integrating point clouds from different sources, while maintaining data integrity in key areas such as building tops, facades, and street surfaces. This ensures both the accuracy and integrity of the point cloud, and effectively smooths the boundaries between point clouds from different sources, thereby improving measurement efficiency. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A flowchart illustrating the multi-source point cloud fusion method for existing buildings provided in an embodiment of the present invention; Figure 2 is a schematic diagram of the registration effect of the multi-source point cloud fusion method for existing buildings provided in the embodiment of the present invention; Figure 3 is a schematic diagram of the screening effect of the multi-source point cloud fusion method for existing buildings provided in the embodiment of the present invention; Figure 4 is a schematic diagram of the fusion effect of the multi-source point cloud fusion method for existing buildings provided in the embodiment of the present invention; Figure 5 A schematic block diagram of an existing building multi-source point cloud fusion device provided in an embodiment of the present invention; Figure 6 This is a schematic block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0015] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0016] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0017] It should be noted that if any AI models, software tools, or components not belonging to the applicant appear in the embodiments of this application, they are merely illustrative examples and do not represent actual use. The user personal information involved in the embodiments of this application is obtained by an entity authorized (knowing and consenting) by the relevant parties or fully authorized by all parties through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0018] This invention provides a method, apparatus, device, and medium for multi-source point cloud fusion of existing buildings. The method is applicable to building surveying, specifically for acquiring comprehensive data of existing buildings, such as for large-scale, high-completeness acquisition for street-level modeling. This invention obtains raw point cloud data by scanning the existing building using multiple scanning devices; registers the raw point cloud data based on a preset registration strategy to obtain first point cloud data; filters the first point cloud data according to a preset filtering strategy to obtain second point cloud data; and fuses the second point cloud data based on a preset fusion strategy to obtain target point cloud data, thereby improving the measurement efficiency of multi-source point cloud fusion of existing buildings. Specific embodiments are described in detail below.

[0019] Please see Figure 1 To Figure 4, Figure 1Figure 1 is a flowchart illustrating the multi-source point cloud fusion method for existing buildings provided in an embodiment of the present invention; Figure 2 is a schematic diagram illustrating the registration effect of the multi-source point cloud fusion method for existing buildings provided in an embodiment of the present invention; Figure 3 is a schematic diagram illustrating the screening effect of the multi-source point cloud fusion method for existing buildings provided in an embodiment of the present invention; Figure 4 is a schematic diagram illustrating the fusion effect of the multi-source point cloud fusion method for existing buildings provided in an embodiment of the present invention. Figure 1 As shown, the method includes the following steps S110-S140.

[0020] S110. Use various scanning devices to scan existing buildings to obtain raw point cloud data.

[0021] In this embodiment, the various scanning devices may include fixed laser scanners, mobile scanners, and drones, etc., to scan the existing building to obtain the original point cloud data. The fixed laser scanner, by deploying multiple scanning stations at key street nodes and inside buildings, uses a fixed ground-based 3D laser scanner to emit laser pulses and receive reflected information, acquiring high-precision 3D coordinate point clouds within the station's field of view. These points are then stitched together to form a complete scene. The mobile laser scanner refers to an operator carrying a backpack or handheld laser scanning device, moving along a planned path within the scene. The device collects environmental point clouds in real time during its movement and uses Simultaneous Localization and Mapping (SLAM) technology to calculate and generate a large-scale, continuous point cloud model. The scanning methods of the drone equipment can include close-up photography and oblique photography. Oblique photography uses a drone equipped with a multi-lens camera to fly over the target street along a pre-set reciprocating flight path, acquiring high-definition images from multiple perspectives. Finally, a large-scale 3D model is generated based on an image 3D modeling algorithm. Close-up photography involves operating the drone close to the building surface, flying along the building's contours to acquire clearer and more overlapping multi-perspective images. These images are then used to generate a high-texture, detailed local point cloud model using an image 3D modeling algorithm.

[0022] In a specific project, the fixed laser scanner can be used to scan the interior spaces of 34 buildings, with a total of 156 stations; the mobile scanner can be used for rapid data acquisition along one street and in 34 interior spaces, with a total length of approximately 400 meters; the drone equipment can be equipped with a multi-lens camera to perform oblique photography and close-up photography tasks; the oblique photography flight altitude can reach 20 meters, with a ground resolution of 0.44 cm / pixel; while the close-up photography mainly captures outdoor facade images at a distance of approximately 1 meter from the facade, with an actual resolution of 0.02 cm / pixel.

[0023] The existing buildings refer to all types of buildings that have been built and put into use, including industrial buildings, civil buildings (including residential buildings and public buildings), and urban infrastructure, such as old neighborhoods and other building complexes.

[0024] S120. Based on a preset registration strategy, the original point cloud data is registered to obtain the first point cloud data.

[0025] In this embodiment, after obtaining the original point cloud data, the original point cloud data can be registered based on a preset registration strategy to obtain the first point cloud data.

[0026] In one embodiment, the process of registering the original point cloud data based on a preset registration strategy to obtain the first point cloud data includes: Obtain first and second original point cloud data of the same area of ​​the existing building from the original point cloud data; For each point in the first original point cloud data and each point in the second original point cloud data, a target nearest neighbor region is obtained by searching the original point cloud data using a preset first algorithm. The target point cloud covariance set is obtained by calculating and processing the target nearest neighbor point domain using a preset second algorithm; The first point cloud data is obtained by performing overlap processing on the covariance set of the target point cloud based on the distribution overlap strategy.

[0027] In this embodiment, first and second original point cloud data of the same area of ​​the existing building are obtained from the original point cloud data. Specifically, for the same area of ​​the existing building, the first and second original point cloud data are obtained from the original point cloud data. The first and second original point cloud data correspond to any two point clouds in the same space, and can be data from two different scanning methods, but not necessarily point clouds from different scanning methods. They can also be point clouds of the same space scanned at different times, depending on the specific scenario.

[0028] For each point in the first original point cloud data and each point in the second original point cloud data, a target nearest neighbor region is obtained by searching the original point cloud data using a preset first algorithm. Specifically, in the original point cloud data, a first nearest neighbor region is obtained by searching each point in the first original point cloud data according to the first algorithm; a second nearest neighbor region is obtained by searching each point in the second original point cloud data according to the first algorithm; the first nearest neighbor region and the second nearest neighbor region are used as the target nearest neighbor region. The first algorithm is a kd-tree (short for k-dimensional tree) search algorithm.

[0029] The target point cloud covariance set is obtained by calculating the target nearest neighbor region using a preset second algorithm. Specifically, the first point cloud covariance set is obtained by calculating the covariance matrix of the first nearest neighbor region in the target nearest neighbor region according to the second algorithm; the second point cloud covariance set is obtained by calculating the covariance matrix of the second nearest neighbor region in the target nearest neighbor region according to the second algorithm; the first point cloud covariance set and the second point cloud covariance set are used as the target point cloud covariance set. The second algorithm is Principal Component Analysis (PCA) algorithm.

[0030] In one embodiment, the step of processing the covariance set of the target point cloud based on a distribution overlap strategy to obtain the first point cloud data includes: The first point cloud data is obtained by iteratively calculating the covariance set of the target point cloud using the constructed objective function and the preset third algorithm.

[0031] In this embodiment, the objective function is a weighted least squares objective function, which has two undetermined coordinate transformation matrices R and t to measure the error between the probability distributions of local shape features described by the two covariance matrices. Then, iterative optimization is performed to minimize the squares objective function. The Gauss-Newton method is applied to iteratively solve for the coordinate transformation matrices R and t, so that the objective function is minimized, that is, the probability distributions of the first and second original point cloud data after coordinate transformation have the highest overlap, thus obtaining the optimal coordinate transformation matrix to complete the registration of the original point cloud data and obtain the first point cloud data.

[0032] The formula corresponding to the objective function is: ,in, p i These are the points in the first original point cloud data.q i It is in the second original point cloud data and p i For the corresponding points, R and t are the transformation matrices to be determined, and C i p And is a point p i The local covariance matrix, C i q They are points q i The local covariance matrix. The preset registration strategy is generalized iterative nearest-point registration.

[0033] In a specific project practice, as shown in Figure 2, the alignment of a fixed scan point cloud (blue) and a moving scan point cloud (yellow) is taken as an example; Figure 2a This is a schematic diagram of the effect before registration provided in an embodiment of the present invention; Figure 2b This is a schematic diagram illustrating the registration effect provided in an embodiment of the present invention; specifically, Figure 2a There are obvious translation and rotation errors between the first two sets of point clouds for registration. However, after applying the preset registration strategy, as shown in the example... Figure 2b The two points are precisely aligned in space, with the average distance between corresponding points being less than 3cm, which meets the accuracy requirements for subsequent fusion.

[0034] As demonstrated by the above embodiments, the first point cloud data is obtained by registering the original point cloud data based on a preset registration strategy. Therefore, it fully considers local geometry, exhibits stronger robustness against noise and outliers, and can better handle point clouds with low overlap or significant density differences. It is particularly suitable for multi-source data registration in complex architectural scenarios, thereby improving measurement efficiency.

[0035] S130. The first point cloud data is filtered and processed according to a preset filtering strategy to obtain the second point cloud data.

[0036] In this embodiment, after obtaining the first point cloud data, the first point cloud data can be filtered according to a preset filtering strategy to obtain the second point cloud data.

[0037] In one embodiment, the step of filtering the first point cloud data according to a preset filtering strategy to obtain the second point cloud data includes: Using the regional scene and data source type of the existing building as selection criteria, a reference point cloud for each region in the existing building is obtained by selecting from the first point cloud data; Obtain the other point clouds in the first point cloud data for each area of ​​the existing building, excluding the reference point cloud, and perform distance calculation processing on the nearest neighbor point of each point in the other point clouds and the reference point cloud according to the distance calculation strategy to obtain a point distance set; The preset distance threshold is compared with the distance of each sub-point in the point distance set; If the distance threshold is greater than or equal to the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are retained in the second point cloud data.

[0038] In this embodiment, the reference point cloud for each area of ​​the existing building is obtained by selecting from the first point cloud data based on the regional scene and data source type of the existing building. Specifically, in the first point cloud data, the regional scene and the data source type are used as selection conditions to obtain the reference point cloud for each area of ​​the existing building, that is, the reference point cloud is the data source with the highest accuracy under the regional scene. The regional scene can include indoor scenes, outdoor scenes, etc. For example, a fixed laser scanning point cloud is used for indoor scenes, while a drone-based close-up photography point cloud is used for outdoor facades. The data source type refers to different point cloud acquisition methods, and different point cloud acquisition methods have theoretical accuracy. For example, the theoretical point cloud accuracy of a fixed laser scanner is 0.1-0.3 cm, the theoretical point cloud accuracy of a backpack-style mobile scanner is 1-3 cm, the theoretical point cloud accuracy of a handheld mobile scanner is 3-5 cm, the theoretical point cloud accuracy of oblique photography by a drone is 5-10 cm, and the theoretical point cloud accuracy of close-up photography by a drone is 3-5 cm. Therefore, based on the above, a priority can be assigned to these five acquisition methods: fixed > backpack > handheld > drone close-up > drone oblique photography. The point cloud with the higher priority will become the baseline point cloud. Additionally, point clouds acquired using different devices and methods can also be used to determine the baseline point cloud based on their relative accuracy. When accuracies are similar, laser point clouds are preferred (as they are active acquisition methods with more stable accuracy).

[0039] The process involves acquiring other point clouds (excluding the reference point cloud) from the first point cloud data for each region of the existing building, and calculating the distance between each point in these other point clouds and its nearest neighbor in the reference point cloud according to a distance calculation strategy to obtain a point distance set. Specifically, for each region of the existing building, the process involves taking other point clouds from the first point cloud data (excluding the reference point cloud), calculating the three-dimensional Euclidean distance from each point in these other point clouds to its nearest neighbor according to a distance calculation strategy, and combining these sub-point distances to obtain the point distance set. It can be seen that a spatial index is established for other point clouds with the reference point cloud as a reference, that is, a reference point cloud is determined in different scenarios, and other point clouds are added to the reference point cloud.

[0040] The step involves comparing a preset distance threshold with the distance of each sub-point in the point distance set. Specifically, the distance threshold is the theoretical point cloud accuracy corresponding to the point cloud acquisition method, so as to compare the distance threshold with the distance of each sub-point.

[0041] If the distance threshold is greater than or equal to the sub-point distance, then the points in the other point clouds corresponding to the sub-point distance are retained in the second point cloud data. Specifically, when the distance threshold is greater than or equal to the sub-point distance, it means that the point is located in the data hole area of ​​the reference point cloud and is retained, and the points in the other point clouds corresponding to the sub-point distance are retained in the second point cloud data.

[0042] In one embodiment, after the step of comparing the preset distance threshold with the distance of each sub-point in the point distance set, the method further includes: If the distance threshold is less than the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are deleted.

[0043] In this embodiment, when the distance threshold is less than the distance to the sub-point, it indicates that the point overlaps with a high-precision reference point and is therefore deleted.

[0044] In a specific project, point cloud filtering refers to retaining only high-precision point clouds in overlapping areas, thereby avoiding the mixing of point clouds with different precision. As shown in Figure 3... Figure 3a The image before filtering the backpack-style point cloud (other point clouds) is more complete, but the point cloud is sparse and the surface texture is poorly expressed. Figure 3b This is a preliminary image of the stand-up point cloud (baseline point cloud) before screening. Although the point cloud is highly accurate and dense, it is missing parts of the surface of the structural columns due to the obstruction of the corners. For example... Figure 3c As shown, if the two types of data are aligned and superimposed, it will result in point cloud clutter and texture inconsistency; for example... Figure 3dAs shown, after applying the preset filtering strategy, high-precision points collected by the stand-up collection are retained in the large overlapping area, while backpack-style supplementary points are retained in the non-overlapping area, resulting in a preliminary fused point cloud that balances accuracy and completeness.

[0045] Meanwhile, in a specific project, mobile laser scanning can efficiently acquire geometric information of streets, building bases, and facades, but it cannot capture the building roof structure. While UAV oblique photography can fully capture the building's top shape, its modeling of the arcade base and facade is poor, with numerous data gaps and geometric distortions. Aligning and overlaying the two types of data, although complementary in form, produces significant data mixing in overlapping areas (such as eaves junctions and outer columns). This study uses mobile laser point clouds as a baseline and UAV point cloud data (along with other point clouds) to accurately fill in the roof gaps. An optimized algorithm smooths the boundary areas, resulting in fused target point cloud data to complete the final fusion model. The final fusion model achieves full-space, full-element coverage, with a model completeness far exceeding any single source, fully demonstrating the necessity of point cloud fusion for macro-level scene modeling.

[0046] Meanwhile, in specific project practice, the facades of old neighborhoods (such as gables, stucco, decorative lines, etc.) carry important historical and cultural information. Their renovation and restoration require high-fidelity recording of fine features and precise positioning within the overall architectural coordinate system. While mobile laser point clouds can provide the overall outline and spatial location of the facade, their sparse point clouds and blurred textures completely fail to meet the needs of detailed modeling of facade details. Although close-up drone photography only provides local information, the reconstructed point cloud has extremely high texture clarity and geometric precision, clearly restoring the decorative details of the facade. After point cloud alignment and overlay, the high-precision local close-up photographic model is positioned within the overall street scene, allowing all facades to be analyzed under a unified coordinate system. Furthermore, after fusion and optimization, the texture inconsistency caused by overlay is avoided. The final model presents high-definition textures from close-up drone photography in facade details, while retaining the complete outline of the mobile point cloud in other areas, achieving a balance between high precision and high integrity. This can serve as a tool for the digital inheritance and comparative verification of facade restoration.

[0047] Meanwhile, in a specific project, the structural safety assessment of existing buildings relies on precise mapping of wall properties such as flatness, verticality, and deformation. This requires point cloud data to simultaneously meet the requirements of millimeter-level high precision and high completeness. While mobile laser point clouds provide complete data, their centimeter-level precision and sparse density result in numerous noise points and large errors in flatness analysis results, rendering them unusable for structural analysis. Fixed laser scanning meets millimeter-level precision requirements, producing smooth and accurate flatness analysis maps. However, due to the obstruction of a wooden beam attached to the wall in this project, a transverse void exists in the middle of the wall, making the analysis results incomplete and questionable in reliability. Although aligning and overlaying the two data sources fills the gaps, the low-precision mobile point cloud mixed with the high-precision data distorts the flatness analysis results. By using the fixed point cloud as the baseline and supplementing and fusing the mobile point cloud only in the void area, millimeter-level precision is maintained for most areas while improving the completeness and reliability of the missing areas. This solves the problem of obstruction and missing data in high-precision mapping in complex urban renewal scenarios.

[0048] As can be seen from the above embodiments, the second point cloud data is obtained by filtering the first point cloud data according to the preset filtering strategy, so as to ensure both the accuracy and integrity of the point cloud, thereby improving the measurement efficiency.

[0049] S140. The second point cloud data is fused based on a preset fusion strategy to obtain the target point cloud data.

[0050] In this embodiment, after obtaining the second point cloud data, the second point cloud data can be fused based on a preset fusion strategy to obtain the target point cloud data.

[0051] In one embodiment, the step of fusing the second point cloud data based on a preset fusion strategy to obtain the target point cloud data includes: The second point cloud data is downsampled to obtain the third point cloud data; Obtain the normal vector data of the third point cloud data, and construct a continuous surface using the normal vector data; Obtain several triangular facets from the continuous surface; The target point cloud data is obtained by sampling and fusing several triangular facets using a sampling and fusion strategy.

[0052] In this embodiment, the second point cloud data is downsampled to obtain the third point cloud data. Specifically, the second point cloud data is voxelized and downsampled according to a preset voxel grid to obtain the third point cloud data. The voxelization and downsampling process is to homogenize the point cloud density and significantly reduce data redundancy. The size of the voxel grid can be set to 0.5 cm.

[0053] The process involves acquiring the normal vector data of the third point cloud data and constructing a continuous surface using this normal vector data. Specifically, the normal vector data of the third point cloud data is acquired to construct a continuous implicit surface as the continuous surface. The normal vector data is a unit vector perpendicular to the local surface of the point cloud, used to characterize the orientation of the surface at that point. The purpose of this step is to first construct a continuous surface at the boundaries between point clouds from different sources, as these boundaries are often broken and discontinuous.

[0054] Specifically, the process of obtaining a plurality of triangular facets from the continuous surface refers to obtaining a plurality of triangular facets from the continuous surface.

[0055] In one embodiment, the target point cloud data is obtained by sampling and fusing several triangular patches using a sampling fusion strategy, including: The total surface area is obtained by summing the areas of several of the triangular facets. The point density is determined based on the preset target spacing; The total number of points is estimated using the total surface area and the point density. The target point cloud data is obtained by random sampling based on a preset model and the total number of points.

[0056] In this embodiment, the total surface area is obtained by summing the areas of several triangular facets. Specifically, the total surface area is obtained by iterating through several triangular facets and summing the areas of each triangular facet. The formula for summation is as follows: , Ti It is the i-th triangular facet.

[0057] The point density is determined based on a preset target spacing. Specifically, ideally, if the points are arranged in a hexagonal tessellation (densely packed) on the plane, the area "occupied" by each point is: d is the preset target spacing; therefore, the point density (number of points per unit area) is: .

[0058] The total number of points is estimated using the total surface area and the point density. Specifically, the estimation formula is as follows: N is the estimated total number of points, ρ is the point density, and A is the total surface area.

[0059] The target point cloud data is obtained by random sampling based on a preset model and the total number of points. Specifically, using the total number of points, random sampling is performed on the preset model according to the area ratio (i.e., more points are sampled for large areas and fewer points are sampled for small areas) to obtain the target point cloud data, ensuring that the overall distribution is consistent with the geometry. The preset model can be a mesh model; the preset fusion strategy is a point cloud completion algorithm based on voxel downsampling and Poisson surface reconstruction.

[0060] After obtaining the target point cloud data, the target point cloud data is used as the final generated fused point cloud model.

[0061] In a specific project, as shown in Figure 4, after point cloud registration and filtering, the density may still vary significantly in different regions, and there is an uneven transition at the boundaries. Further optimization using voxelization downsampling and surface sampling fusion reconstruction methods is necessary. Figure 4a The image shows the preliminary fusion result of two point clouds, which is a schematic diagram of the filtered point cloud effect; as shown... Figure 4b The image shown is a schematic diagram illustrating the effect of point cloud downsampling; as shown... Figure 4c As shown, this is a schematic diagram of the effect of constructing a continuous implicit surface based on the normal vector information of the point cloud, and then resampling from this continuous surface to obtain the optimized point cloud. This is the effect diagram of the sampling fusion result of continuous point cloud interpolation.

[0062] As can be seen from the above embodiments, the target point cloud data obtained by fusing the second point cloud data based on the preset fusion strategy can effectively smooth the boundaries between point clouds from different sources, fill in tiny holes to greatly reduce visual abruptness, ensure the overall quality of the final fused point cloud, and has high engineering applicability and feasibility. It successfully meets the application requirements of three levels: street-level overall modeling, detailed restoration of facade features, and mapping of existing wall structure attributes, proving the necessity and superiority of the multi-source fusion path, thereby improving measurement efficiency.

[0063] As can be seen, to address the diverse needs of old neighborhood renovation projects for 3D models in terms of multi-scale, multi-precision, and texture detail, this paper constructs a multi-source heterogeneous point cloud technology framework encompassing multi-source acquisition, quantitative evaluation, alignment and fusion, and layered application. This framework aims to overcome the inherent limitations of single acquisition technologies and generate unified digital models with high precision, high integrity, and high-quality texture.

[0064] Engineering practice based on the above-mentioned project practice shows that: (1) the proposed technical framework has high engineering applicability and feasibility; (2) the fusion method based on generalized iterative nearest point (G-ICP) registration and spatial constraint optimization can effectively integrate the technical advantages of point clouds from different sources, and significantly improve the data integrity of the model in key areas while retaining high-precision data and high-definition textures; (3) the final fused point cloud model successfully meets the application requirements of three levels: street-level overall modeling, detailed restoration of facade features, and mapping of existing wall structure attributes, which confirms the necessity and superiority of the multi-source fusion path.

[0065] While this invention has achieved its intended results, there is still room for optimization. Firstly, the selection of the baseline point cloud and the setting of the screening threshold in the current fusion strategy rely to some extent on engineering experience. Future research could explore adaptive fusion algorithms based on point cloud quality metrics to improve the scenario generalization of the fusion process. Furthermore, deeper integration of fused point clouds with semantic information could be further investigated to achieve automated segmentation and component recognition of 3D models of old urban blocks, providing richer data support for subsequent building information model delivery and intelligent operation and maintenance.

[0066] In summary, the embodiments of the present invention can utilize multiple scanning devices to scan existing buildings to obtain raw point cloud data; perform registration processing on the raw point cloud data based on a preset registration strategy to obtain first point cloud data; perform filtering processing on the first point cloud data according to a preset filtering strategy to obtain second point cloud data; and perform fusion processing on the second point cloud data based on a preset fusion strategy to obtain target point cloud data. Therefore, the embodiments of the present invention utilize the aforementioned multiple scanning devices to scan the existing buildings to obtain the raw point cloud data, and perform registration, filtering, and fusion processing on the raw point cloud data to obtain the target point cloud data. This approach can handle point clouds with low overlap or significant density differences, effectively integrate the technical advantages of point clouds from different sources, retain high-precision data and high-definition textures while ensuring the data integrity of key areas such as building tops, facades, and street surfaces, guaranteeing both the accuracy and integrity of the point cloud, and effectively smoothing the boundaries between point clouds from different sources, thereby improving measurement efficiency.

[0067] Please see Figure 5 , Figure 5 This is a schematic block diagram of an existing building multi-source point cloud fusion device provided in an embodiment of the present invention. Figure 5 As shown, corresponding to the above-mentioned method for multi-source point cloud fusion of existing buildings, this invention also provides a device for multi-source point cloud fusion of existing buildings. This device is suitable for applications in the field of building surveying, specifically for the measurement and acquisition of overall data of existing buildings, such as the need for large-scale, high-completeness acquisition of street-level modeling. For details, please refer to... Figure 5The existing building multi-source point cloud fusion device 700 includes: The scanning unit 701 is used to scan existing buildings using various scanning devices to obtain raw point cloud data; Registration unit 702 is used to perform registration processing on the original point cloud data based on a preset registration strategy to obtain first point cloud data; The filtering unit 703 is used to filter the first point cloud data according to a preset filtering strategy to obtain the second point cloud data. The fusion unit 704 is used to perform fusion processing on the second point cloud data based on a preset fusion strategy to obtain the target point cloud data.

[0068] In some embodiments, when the registration unit 702 performs the step of registering the original point cloud data based on a preset registration strategy to obtain the first point cloud data, it is specifically used for: Obtain first and second original point cloud data of the same area of ​​the existing building from the original point cloud data; For each point in the first original point cloud data and each point in the second original point cloud data, a target nearest neighbor region is obtained by searching the original point cloud data using a preset first algorithm. The target point cloud covariance set is obtained by calculating and processing the target nearest neighbor point domain using a preset second algorithm; The first point cloud data is obtained by performing overlap processing on the covariance set of the target point cloud based on the distribution overlap strategy.

[0069] In some embodiments, when the filtering unit 703 performs the processing step of filtering the first point cloud data according to a preset filtering strategy to obtain the second point cloud data, it is specifically used for: Using the regional scene and data source type of the existing building as selection criteria, a reference point cloud for each region in the existing building is obtained by selecting from the first point cloud data; Obtain the other point clouds in the first point cloud data for each area of ​​the existing building, excluding the reference point cloud, and perform distance calculation processing on the nearest neighbor point of each point in the other point clouds and the reference point cloud according to the distance calculation strategy to obtain a point distance set; The preset distance threshold is compared with the distance of each sub-point in the point distance set; If the distance threshold is greater than or equal to the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are retained in the second point cloud data.

[0070] In some embodiments, when the fusion unit 704 performs the processing step of fusing the second point cloud data based on a preset fusion strategy to obtain the target point cloud data, it is specifically used for: The second point cloud data is downsampled to obtain the third point cloud data; Obtain the normal vector data of the third point cloud data, and construct a continuous surface using the normal vector data; Obtain several triangular facets from the continuous surface; The target point cloud data is obtained by sampling and fusing several triangular facets using a sampling and fusion strategy.

[0071] In some embodiments, when performing the processing step of sampling and fusing several of the triangular facets using a sampling fusion strategy to obtain the target point cloud data, the specific steps are as follows: The total surface area is obtained by summing the areas of several of the triangular facets. The point density is determined based on the preset target spacing; The total number of points is estimated using the total surface area and the point density. The target point cloud data is obtained by random sampling based on a preset model and the total number of points.

[0072] In some embodiments, after performing the step of comparing the preset distance threshold with the distance of each sub-point in the point distance set, the method is further configured to: If the distance threshold is less than the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are deleted.

[0073] In some embodiments, when performing the processing step of processing the covariance set of the target point cloud based on the distribution overlap strategy to obtain the first point cloud data, the specific steps are as follows: The first point cloud data is obtained by iteratively calculating the covariance set of the target point cloud using the constructed objective function and the preset third algorithm.

[0074] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned existing building multi-source point cloud fusion device and each unit can be referred to the corresponding description in the foregoing method embodiments. For the sake of convenience and brevity, it will not be repeated here.

[0075] The aforementioned multi-source point cloud fusion device for existing buildings can be implemented as a computer program, which can, for example... Figure 6 It runs on the electronic device shown.

[0076] Please see Figure 6 , Figure 6 This is a schematic block diagram of an electronic device provided in an embodiment of the present invention. The electronic device 800 can be a terminal or a server. The terminal can be an electronic device with communication functions. The server can be a standalone server or a server cluster composed of multiple servers.

[0077] See Figure 6 The electronic device 800 includes a processor 802, a memory, and a network interface 805 connected via a system bus 801. The memory may include a non-volatile storage medium 803 and internal memory 804.

[0078] The non-volatile storage medium 803 may store an operating system 8031 ​​and a computer program 8032. The computer program 8032 includes program instructions that, when executed, cause the processor 802 to perform a method for fusing multi-source point clouds of existing buildings.

[0079] The processor 802 provides computing and control capabilities to support the operation of the entire electronic device 800.

[0080] The internal memory 804 provides an environment for the operation of the computer program 8032 in the non-volatile storage medium 803. When the computer program 8032 is executed by the processor 802, the processor 802 can execute a method for fusing multi-source point clouds of existing buildings.

[0081] This network interface 805 is used for network communication with other devices. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the electronic device 800 to which the present invention is applied. The specific electronic device 800 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0082] The processor 802 is used to run a computer program 8032 stored in the memory to perform the following steps: Raw point cloud data was obtained by scanning existing buildings using various scanning devices; The original point cloud data is registered based on a preset registration strategy to obtain the first point cloud data. The first point cloud data is filtered and processed according to a preset filtering strategy to obtain the second point cloud data; The target point cloud data is obtained by fusing the second point cloud data based on a preset fusion strategy.

[0083] In some embodiments, when implementing the step of registering the original point cloud data based on a preset registration strategy to obtain the first point cloud data, the processor 802 is specifically used for: Obtain first and second original point cloud data of the same area of ​​the existing building from the original point cloud data; For each point in the first original point cloud data and each point in the second original point cloud data, a target nearest neighbor region is obtained by searching the original point cloud data using a preset first algorithm. The target point cloud covariance set is obtained by calculating and processing the target nearest neighbor point domain using a preset second algorithm; The first point cloud data is obtained by performing overlap processing on the covariance set of the target point cloud based on the distribution overlap strategy.

[0084] In some embodiments, when implementing the step of filtering the first point cloud data according to a preset filtering strategy to obtain the second point cloud data, the processor 802 is specifically used for: Using the regional scene and data source type of the existing building as selection criteria, a reference point cloud for each region in the existing building is obtained by selecting from the first point cloud data; Obtain the other point clouds in the first point cloud data for each area of ​​the existing building, excluding the reference point cloud, and perform distance calculation processing on the nearest neighbor point of each point in the other point clouds and the reference point cloud according to the distance calculation strategy to obtain a point distance set; The preset distance threshold is compared with the distance of each sub-point in the point distance set; If the distance threshold is greater than or equal to the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are retained in the second point cloud data.

[0085] In some embodiments, when implementing the processing step of fusing the second point cloud data based on a preset fusion strategy to obtain target point cloud data, the processor 802 is specifically used for: The second point cloud data is downsampled to obtain the third point cloud data; Obtain the normal vector data of the third point cloud data, and construct a continuous surface using the normal vector data; Obtain several triangular facets from the continuous surface; The target point cloud data is obtained by sampling and fusing several triangular facets using a sampling and fusion strategy.

[0086] In some embodiments, when implementing the processing step of sampling and fusing several triangular facets using a sampling fusion strategy to obtain the target point cloud data, the processor 802 is specifically used for: The total surface area is obtained by summing the areas of several of the triangular facets. The point density is determined based on the preset target spacing; The total number of points is estimated using the total surface area and the point density. The target point cloud data is obtained by random sampling based on a preset model and the total number of points.

[0087] In some embodiments, after implementing the processing step of comparing the preset distance threshold with the distance of each sub-point in the point distance set, the processor 802 is further configured to: If the distance threshold is less than the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are deleted.

[0088] In some embodiments, when implementing the processing step of performing overlap degree processing on the target point cloud covariance set based on the distribution overlap strategy to obtain the first point cloud data, the processor 802 is specifically used for: The first point cloud data is obtained by iteratively calculating the covariance set of the target point cloud using the constructed objective function and the preset third algorithm.

[0089] It should be understood that, in this embodiment of the invention, the processor 802 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0090] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0091] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When executed by a processor, the program instructions cause the processor to perform the following steps: Raw point cloud data was obtained by scanning existing buildings using various scanning devices; The original point cloud data is registered based on a preset registration strategy to obtain the first point cloud data. The first point cloud data is filtered and processed according to a preset filtering strategy to obtain the second point cloud data; The target point cloud data is obtained by fusing the second point cloud data based on a preset fusion strategy.

[0092] In one embodiment, when the processor executes the program instructions to implement the processing step of registering the original point cloud data based on a preset registration strategy to obtain the first point cloud data, it is specifically used for: Obtain first and second original point cloud data of the same area of ​​the existing building from the original point cloud data; For each point in the first original point cloud data and each point in the second original point cloud data, a target nearest neighbor region is obtained by searching the original point cloud data using a preset first algorithm. The target point cloud covariance set is obtained by calculating and processing the target nearest neighbor point domain using a preset second algorithm; The first point cloud data is obtained by performing overlap processing on the covariance set of the target point cloud based on the distribution overlap strategy.

[0093] In one embodiment, when the processor executes the program instructions to implement the processing step of filtering the first point cloud data according to a preset filtering strategy to obtain the second point cloud data, it is specifically used for: Using the regional scene and data source type of the existing building as selection criteria, a reference point cloud for each region in the existing building is obtained by selecting from the first point cloud data; Obtain the other point clouds in the first point cloud data for each area of ​​the existing building, excluding the reference point cloud, and perform distance calculation processing on the nearest neighbor point of each point in the other point clouds and the reference point cloud according to the distance calculation strategy to obtain a point distance set; The preset distance threshold is compared with the distance of each sub-point in the point distance set; If the distance threshold is greater than or equal to the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are retained in the second point cloud data.

[0094] In one embodiment, when the processor executes the program instructions to implement the processing step of fusing the second point cloud data based on a preset fusion strategy to obtain the target point cloud data, it is specifically used for: The second point cloud data is downsampled to obtain the third point cloud data; Obtain the normal vector data of the third point cloud data, and construct a continuous surface using the normal vector data; Obtain several triangular facets from the continuous surface; The target point cloud data is obtained by sampling and fusing several triangular facets using a sampling and fusion strategy.

[0095] In one embodiment, when the processor executes the program instructions to implement the processing step of sampling and fusing several triangular facets using a sampling fusion strategy to obtain the target point cloud data, it is specifically used for: The total surface area is obtained by summing the areas of several of the triangular facets. The point density is determined based on the preset target spacing; The total number of points is estimated using the total surface area and the point density. The target point cloud data is obtained by random sampling based on a preset model and the total number of points.

[0096] In one embodiment, after executing the program instructions to implement the processing step of comparing the preset distance threshold with the distance of each sub-point in the point distance set, the processor is further configured to: If the distance threshold is less than the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are deleted.

[0097] In one embodiment, when the processor executes the program instructions to implement the processing step of processing the overlap degree of the target point cloud covariance set based on the distribution overlap strategy to obtain the first point cloud data, it is specifically used for: The first point cloud data is obtained by iteratively calculating the covariance set of the target point cloud using the constructed objective function and the preset third algorithm.

[0098] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0099] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0100] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0101] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0102] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0103] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for fusing multi-source point clouds of existing buildings, characterized in that, The existing building multi-source point cloud fusion method includes: Raw point cloud data was obtained by scanning existing buildings using various scanning devices; The original point cloud data is registered based on a preset registration strategy to obtain the first point cloud data. The first point cloud data is filtered and processed according to a preset filtering strategy to obtain the second point cloud data; The target point cloud data is obtained by fusing the second point cloud data based on a preset fusion strategy.

2. The method for multi-source point cloud fusion of existing buildings according to claim 1, characterized in that, The process of registering the original point cloud data based on a preset registration strategy to obtain the first point cloud data includes: Obtain first and second original point cloud data of the same area of ​​the existing building from the original point cloud data; For each point in the first original point cloud data and each point in the second original point cloud data, a target nearest neighbor region is obtained by searching the original point cloud data using a preset first algorithm. The target point cloud covariance set is obtained by calculating and processing the target nearest neighbor point domain using a preset second algorithm; The first point cloud data is obtained by performing overlap processing on the covariance set of the target point cloud based on the distribution overlap strategy.

3. The method for multi-source point cloud fusion of existing buildings according to claim 1, characterized in that, The step of filtering the first point cloud data according to a preset filtering strategy to obtain the second point cloud data includes: Using the regional scene and data source type of the existing building as selection criteria, a reference point cloud for each region in the existing building is obtained by selecting from the first point cloud data; Obtain the other point clouds in the first point cloud data for each area of ​​the existing building, excluding the reference point cloud, and perform distance calculation processing on the nearest neighbor point of each point in the other point clouds and the reference point cloud according to the distance calculation strategy to obtain a point distance set; The preset distance threshold is compared with the distance of each sub-point in the point distance set; If the distance threshold is greater than or equal to the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are retained in the second point cloud data.

4. The method for multi-source point cloud fusion of existing buildings according to claim 1, characterized in that, The process of fusing the second point cloud data based on a preset fusion strategy to obtain the target point cloud data includes: The second point cloud data is downsampled to obtain the third point cloud data; Obtain the normal vector data of the third point cloud data, and construct a continuous surface using the normal vector data; Obtain several triangular facets from the continuous surface; The target point cloud data is obtained by sampling and fusing several triangular facets using a sampling and fusion strategy.

5. The method for multi-source point cloud fusion of existing buildings according to claim 4, characterized in that, The step of using a sampling fusion strategy to sample and fuse several of the triangular facets to obtain the target point cloud data includes: The total surface area is obtained by summing the areas of several of the triangular facets. The point density is determined based on the preset target spacing; The total number of points is estimated using the total surface area and the point density. The target point cloud data is obtained by random sampling based on a preset model and the total number of points.

6. The method for multi-source point cloud fusion of existing buildings according to claim 3, characterized in that, After the step of comparing the preset distance threshold with the distance of each sub-point in the point distance set, the method further includes: If the distance threshold is less than the sub-point distance, then the points in the other point cloud corresponding to the sub-point distance are deleted.

7. The method for multi-source point cloud fusion of existing buildings according to claim 2, characterized in that, The first point cloud data is obtained by performing overlap processing on the target point cloud covariance set based on the distribution overlap strategy, including: The first point cloud data is obtained by iteratively calculating the covariance set of the target point cloud using the constructed objective function and the preset third algorithm.

8. A multi-source point cloud fusion device for existing buildings, characterized in that, The existing building multi-source point cloud fusion device includes: The scanning unit is used to scan existing buildings using various scanning devices to obtain raw point cloud data; A registration unit is used to perform registration processing on the original point cloud data based on a preset registration strategy to obtain the first point cloud data. The filtering unit is used to filter the first point cloud data according to a preset filtering strategy to obtain the second point cloud data; The fusion unit is used to perform fusion processing on the second point cloud data based on a preset fusion strategy to obtain the target point cloud data.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the existing building multi-source point cloud fusion method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which includes program instructions that, when executed by a processor, cause the processor to perform the existing building multi-source point cloud fusion method as described in any one of claims 1-7.

Images