Method for Classifying and Recognizing Point Clouds of Building Roofs and Walls Based on UAV Technology

Abstract

The present invention relates to the technical field of point cloud recognition, and discloses a method for classifying and recognizing building roof and wall point clouds based on UAV technology, which includes the following steps: Step 1, obtaining building point cloud data by means of UAV scanning; Step 2, constructing a multi-scale supervoxel fusion model according to the point cloud data obtained in Step 1 and fusing the point clouds; Step 3, recognizing the building point clouds of different surfaces according to the point clouds fused by the multi-scale supervoxel fusion model in Step 2. The present invention focuses on using normal vectors, heights, point cloud densities, and the number of points to realize the recognition of UAV building point cloud walls and roof surfaces, thereby avoiding the problem that wall point clouds are too sparse to be recognized, significantly improving the recognition efficiency and refinement degree of UAV building point clouds, and being applicable to the extraction of walls and roof surfaces of large-scale and a large number of building point clouds collected by UAVs.

Description

Technical Field

[0001] The present invention relates to the technical field of point cloud recognition, and particularly to a method for classifying and recognizing point clouds on building roofs and walls based on UAV technology. Background Art

[0002] With the rapid development of UAV technology and lidar technology, using a UAV (unmanned aerial vehicle, i.e., drone) equipped with lidar for high-precision building point cloud acquisition has become an important means in the fields of building information modeling, urban planning, and intelligent buildings. Point cloud data, with its high resolution and accuracy, provides strong support for three-dimensional building reconstruction. However, due to the characteristics of high density, large scale, and complex distribution of the point cloud data collected by UAVs, and the significant difference in the quality of point clouds on building roofs and walls, traditional building point cloud recognition methods are not applicable, resulting in unsatisfactory recognition of building roofs and walls.

[0003] Existing methods for recognizing UAV building point cloud roofs and walls mainly rely on the following technologies: one is the plane recognition method based on RANSAC, which is applicable to buildings with simple structures but is easily affected by point cloud quality and recognition accuracy when dealing with complex building structures, resulting in insufficient model accuracy; the second is the region-growing-based method, although such methods can handle complex structures, they involve many parameters, and different parameters have a great impact on the recognition results; the third is the supervoxel-based method, which has a high recognition accuracy but currently cannot be applied to the recognition of UAV wall point clouds.

[0004] To address the above technical problems, there is an urgent need for an efficient, robust, and UAV technology-based method for classifying and recognizing point clouds on building roofs and walls that is applicable to complex building structures. Summary of the Invention

[0005] To solve the technical problems raised in the background art, the present invention provides a method for classifying and recognizing point clouds on building roofs and walls based on UAV technology. This method should be able to make full use of building height information, roof and wall direction information, and roof point cloud density information to achieve accurate recognition of complex building roofs and walls, and at the same time possess high efficiency and robustness to meet the rapid processing requirements of large-scale point cloud data.

[0006] The present invention is implemented by the following technical solutions: A method for classifying and recognizing point clouds on building roofs and walls based on UAV technology, comprising the following steps:

[0007] Step 1, obtaining building point cloud data by means of UAV scanning;

[0008] Step 2, constructing a multi-scale supervoxel fusion model based on the point cloud data obtained in Step 1 and fusing the point clouds;

[0009] Step 3: Identify the building point clouds of different faces based on the point clouds fused by the multi-scale supervoxel fusion model in Step 2;

[0010] Step 4: Calculate the normal vectors of the building point clouds of different faces identified in Step 3; and identify the wall point clouds based on the normal information;

[0011] Step 5: After removing the wall point cloud data in Step 4, obtain the point cloud data of the remaining identified faces, and construct a probability model of height, density, and unit point count for the point cloud data of the remaining identified faces. Finally, extract the building roof point clouds from the point cloud data of the remaining identified faces.

[0012] Specifically, the construction operation process of the multi-scale supervoxel fusion model in Step 2 is as follows:

[0013] Step 2.1: Construct supervoxels for the UAV building point cloud data in Step 1 to obtain the initial supervoxel block point cloud, which includes planar supervoxels and non-planar supervoxels;

[0014] Step 2.2: Treat the planar supervoxels and non-planar supervoxels as the basic units of the region growing algorithm, and construct a multi-scale supervoxel fusion algorithm.

[0015] Specifically, the steps for constructing the multi-scale supervoxel fusion algorithm are as follows:

[0016] Step 2.21: Determine the normal vectors of the supervoxel block point cloud; for each point in the supervoxel block point cloud, select its k nearest neighbor points, and the k nearest neighbor points form the local neighborhood of point . Then calculate the mean value of the points in the local neighborhood, as shown in Equation (1):

[0017] (1)

[0018] According to Equation (1), construct the covariance matrix C , as shown in Equation (2):

[0019] (2)

[0020] where are the point coordinates in the neighborhood, and is the average coordinate of these points;

[0021] Perform eigenvalue decomposition on the covariance matrix C to obtain the eigenvalues and their corresponding eigenvectors . The eigenvector corresponding to the minimum eigenvalue That is the estimated normal vector; calculate the normal vectors of all points in the initial supervoxel block The average value is the point cloud normal vector of the supervoxel block. To ensure the consistency of the point cloud normal vector of the supervoxel block, it is necessary to orient the normal vector and select a reference direction , such as the viewing direction or the external normal vector, and check the normal vector and the reference direction 's included angle. If the included angle is greater than 90 degrees, reverse the normal vector, as shown in Equation (3):

[0022] (3)

[0023] Then normalize the normal vector to make it unitized, as shown in Equation (4):

[0024] (4)

[0025] Step 2.22, Calculate the curvature of the supervoxel block point cloud; Curvature is a quantity that measures the degree of bending of the supervoxel surface. According to the three eigenvalues obtained from the eigenvalue decomposition in Equation (2), get the curvature of each point in the supervoxel block point cloud, as shown in Equation (5);

[0026] (5)

[0027] Calculate the curvatures of all points in the supervoxel block point cloud 's average value, and finally get the curvature of the supervoxel block point cloud ;

[0028] Step 2.23, Select the seed supervoxel; Sort all supervoxel blocks in ascending order according to the curvature value, and select the point cloud of the supervoxel block with the smallest curvature as the seed supervoxel. The point cloud of the supervoxel block with a smaller curvature usually indicates that the supervoxel surface is relatively flat and is suitable as the starting point for growth, and initialize the recognition result and the set of visited points;

[0029] Step 2.24, Growth process: Starting from the seed supervoxel, sequentially search for its nearest neighbor supervoxels and grow according to the distance threshold and the normal angle threshold; If the growth condition is met, add the neighboring supervoxels to the seed set and the recognition result, and mark them as visited, so as to obtain the recognition result of the planar supervoxel ;

[0030] Step 2.25, For non-planar supervoxels , obtain the curvature and the normal vector of each non-planar supervoxel, calculate the curvature of the non-planar supervoxel and the curvature Difference value;

[0031] (6)

[0032] Calculate the normal vector of non - planar supervoxels And the angle between the normal vector of each planar supervoxel in region growing Angle;

[0033] (7)

[0034] Simultaneously calculate the distance from each non - planar supervoxel to the nearest point of each planar supervoxel recognition data set ;

[0035] According to formula (6), formula (7) and the distance , obtain the fusion judgment parameter of non - planar supervoxels and region - growing planar recognition supervoxels, as shown in formula (8):

[0036] (8)

[0037] Where: Is the curvature difference between the non - planar supervoxel and the neighboring planar supervoxel, Is the difference value of the normal vector angle between the non - planar supervoxel and the neighboring planar supervoxel, Is the nearest distance between the non - planar supervoxel and the neighboring planar supervoxel;

[0038] Step 2.26, Fusion result: When the distance from each non - planar supervoxel to the planar supervoxel recognition data set Is the smallest, then fuse it into the supervoxel block point cloud of the same surface.

[0039] Specifically, the operation of step 2.26 is as follows: When the planar supervoxel is , the Neighboring non - planar supervoxels of this planar supervoxel are ;

[0040] According to formula (8), calculate the parameter Between the Neighboring non - planar supervoxels and the planar supervoxel ; ;

[0041] When Is the smallest, fuse the non - planar supervoxel And the planar supervoxel Together.

[0042] Specifically, in step 4, for the surface recognition point cloud, the wall surface point cloud is recognized based on the normal information of the super-voxel block point cloud after fusion. When the normal direction is parallel to the xoy plane, it is determined as the wall surface point cloud.

[0043] Specifically, the operation of step 5 is as follows:

[0044] After removing the wall surface point cloud, the remaining recognition point cloud is obtained , and the number of points within each enclosing sphere with a radius of is determined using the enclosing sphere algorithm , and the density of each surface recognition point cloud is obtained , as shown in Equation (9).

[0045] (9)

[0046] where is the number of points of each recognition surface.

[0047] According to Equation (9), its normalized result is obtained, as shown in Equation (10):

[0048] (10)

[0049] Meanwhile, the average height of the point cloud of each recognition surface is calculated, as shown in Equation (11):

[0050] (11)

[0051] According to Equation (11), its normalized result is obtained, as shown in Equation (12):

[0052] (12)

[0053] The number of points of each recognition surface is normalized, as shown in Equation (13):

[0054] (13)

[0055] According to Equations (10), (12), and (13), the probability for extracting the building roof point cloud is obtained, as shown in Equation (14):

[0056] (14)

[0057] The building roof point cloud is extracted from the remaining recognition point cloud using the probability of Equation (14).

[0058] Specifically, when the probability is met, it is recognized as the building roof point cloud.

[0059] The present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the method for classifying and identifying the point clouds of building roofs and walls based on UAV technology as described above is implemented.

[0060] A computer-readable storage medium includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the method for classifying and identifying the point clouds of building roofs and walls based on UAV technology as described above.

[0061] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0062] The method in this paper can quickly identify building point clouds, and the identified building surface point clouds are more detailed. The identification method proposed by the present invention is more robust and is not easily affected by multiple building surfaces. Therefore, the method proposed by the present invention has a better refinement degree for building point cloud identification than the RANSAC method and the region growing method.

[0063] The present invention focuses on using the normal vector, height, point cloud density, and number of points to identify the walls and roof surfaces of building point clouds, thus avoiding the problem that the wall point clouds are too sparse to be identified. The present invention performs roof and wall point cloud identification on the building point clouds obtained based on UAV technology, without involving redundant algorithms and programs, significantly improving the identification efficiency and refinement degree of UAV building point clouds, and thus being applicable to the extraction of the walls and roof surfaces of large-scale and large-quantity building point clouds collected by UAVs. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a principle flow chart of the method for classifying and identifying the point clouds of building roofs and walls proposed by the present invention;

[0065] Figure 2 A comparison diagram of the UAV building roof point clouds extracted by the identification method proposed by the present invention, the RANSAC method, and the region growing method;

[0066] Figure 3 It is a point cloud diagram of the building roof extracted by the identification method proposed by the present invention;

[0067] Figure 4 It is a point cloud diagram of the walls and roof in the building point clouds extracted by the identification method proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0068] Next, in combination with the drawings and specific embodiments, the present invention will be further described. It should be noted that, on the premise of no conflict, the following-described embodiments or technical features can be combined arbitrarily to form new embodiments.

[0069] Example

[0070] Please combine with Figures 1 - 4 , the method for classifying and recognizing building roof and wall point clouds based on UAV technology proposed in this example includes the following steps:

[0071] Step 1, obtain building point cloud data by using UAV scanning;

[0072] Step 2, construct a multi-scale supervoxel fusion model according to the point cloud data obtained in Step 1, and fuse the point cloud;

[0073] Step 3, recognize the building point clouds of different surfaces according to the point cloud fused by the multi-scale supervoxel fusion model in Step 2;

[0074] Step 4, calculate the normal of the point clouds of different surfaces recognized in Step 3; and realize the recognition of the wall

[0075] point cloud according to the normal information;

[0076] Step 5, obtain the remaining recognized point cloud after removing the wall point cloud in Step 4;

[0077] Step 6, construct a probability model of height, density and unit number of points for the remaining recognized point cloud in Step 5.

[0078] It should be noted that:

[0079] Construct supervoxels for the UAV building point cloud data in Step 1 to obtain the initial supervoxel block point cloud, which includes planar supervoxels and non-planar supervoxels. Regard the planar supervoxels and non-planar supervoxels as the basic units of the region growing algorithm. Construct a multi-scale supervoxel fusion algorithm, and the specific steps are as follows:

[0080] Determine the normal vector of the supervoxel block point cloud: For each point in the supervoxel block point cloud, select its k nearest neighbor points. These points form the local neighborhood of the point . Then calculate the mean value of the points in the local neighborhood , as shown in Equation (1).

[0081] (1)

[0082] According to Equation (1), construct the covariance matrix C , as shown in Equation (2):

[0083] (2)

[0084] Among them, is the point coordinate in the neighborhood, It is the average coordinate of these points.

[0085] Perform eigenvalue decomposition on the covariance matrix C to obtain eigenvalues and their corresponding eigenvectors . The eigenvector corresponding to the smallest eigenvalue is the estimated normal vector.

[0086] Calculate the average value of the normal vectors of all points in the initial supervoxel block . The average value is the normal vector of the supervoxel block point cloud. To ensure the consistency of the normal vector of the supervoxel block point cloud, it is necessary to orient the normal vector. Select a reference direction , such as the viewing direction or the external normal vector. Check the angle between the normal vector and the reference direction . If the angle is greater than 90 degrees, reverse the normal vector, as shown in Equation (3):

[0087] (3)

[0088] Then normalize the normal vector to make it unitized, as shown in Equation (4):

[0089] (4)

[0090] Calculation of the curvature of the supervoxel block point cloud: Curvature is a quantity that measures the degree of bending of the supervoxel surface. Based on the three eigenvalues obtained from the eigenvalue decomposition in Equation (2), the curvature of each point in the supervoxel block point cloud is obtained, as shown in Equation (5):

[0091] (5)

[0092] Calculate the average value of the curvatures of all points in the supervoxel block point cloud . Finally, the curvature of the supervoxel block point cloud is obtained .

[0093] Selection of seed supervoxels: Sort all supervoxel blocks in ascending order according to the curvature value, and select the supervoxel block point cloud with the smallest curvature as the seed supervoxel. The supervoxel block point cloud with a smaller curvature usually indicates that the supervoxel surface is relatively flat, which is suitable as the starting point for growth, and initialize the recognition result and the set of visited points.

[0094] Growth process: Starting from the seed supervoxel, search for its nearest neighbor supervoxels in turn, and grow according to the distance threshold and the normal angle threshold. If the growth condition is met, add the neighboring supervoxels to the seed set and the recognition result, and mark them as visited. Thus, the recognition result of the planar supervoxel is obtained .

[0095] For non-planar supervoxels , obtain the curvature and normal vector of each non-planar supervoxel . Calculate the difference between the curvature of the non-planar supervoxel and the curvature of each planar supervoxel in region growing :

[0096]

[0097] Calculate the angle between the normal vector of the non-planar supervoxel and the normal vector of each planar supervoxel in region growing :

[0098]

[0099] At the same time, calculate the distance from each non-planar supervoxel to the nearest point of each planar supervoxel recognition data set .

[0100] According to equations (6), (7) and the distance , obtain the fusion judgment parameter between the non-planar supervoxel and the region growing planar recognition supervoxel, as shown in equation (8):

[0101]

[0102] where: is the curvature difference between the non-planar supervoxel and the neighboring planar supervoxel, is the difference in the angle between the normal vectors of the non-planar supervoxel and the neighboring planar supervoxel, is the nearest distance between the non-planar supervoxel and the neighboring planar supervoxel.

[0103] Fusion result: When the distance from each non-planar supervoxel to the planar supervoxel recognition data set is the smallest, then fuse it, and it is the point cloud of the supervoxel block on the same surface.

[0104] For example, if the planar supervoxel is , and the neighboring non-planar supervoxels of this planar supervoxel are . According to equation (8), calculate the parameter between the neighboring non-planar supervoxels and the planar supervoxel . When is the smallest, fuse the non-planar supervoxel and the planar supervoxel together.

[0105] Thus, surface recognition of the building point cloud is achieved based on the above fusion results.

[0106] Compare the building surface segmentation results obtained by this solution with the RANSAC method and the region growing method, as Figure 2 shown. From Figure 2 the comparison between the RANSAC method in part (a) and Figure 2 this solution in part (c) in

[0107] it can be seen that the wall surface of this building point cloud has been successfully segmented into three small wall surfaces by the recognition method proposed in this solution, while the three small wall surfaces cannot be segmented by using the RANSAC method. Figure 2 From Figure 2 the comparison between the region growing method in part (b) and

[0108] this solution in part (c) in Figure 2 it can be seen that the region growing method not only fails to segment the roof on the right side of this building, but also produces a relatively chaotic segmentation result.

[0109] In this solution, for the surface recognition point cloud, the wall surface recognition point cloud can be determined according to the normal direction of the supervoxel block point cloud after fusion, that is, when the normal direction is parallel to the xoy plane, it is judged as the wall surface point cloud.

[0110] After removing the wall surface point cloud, the remaining recognition point cloud is obtained.

[0111] For the remaining recognition point cloud , the number of points within each enclosing sphere with a radius of is determined by using the enclosing sphere algorithm .

[0112] The density of each surface recognition point cloud is obtained, as shown in Equation (9).

[0113] (9)

[0114] where is the number of points of each recognition surface.

[0115] According to Equation (9), its normalized result is obtained, as shown in Equation (10):

[0116] (10)

[0117] At the same time, the average height of each recognition point cloud is calculated, as shown in Equation (11):

[0118] (11)

[0119] According to formula (11), its normalized result is obtained as shown in formula (12):

[0120] (12)

[0121] Normalize the number of points on each recognized surface as shown in formula (13):

[0122] (13)

[0123] According to formula (10), (12) and (13), the probability for extracting the building roof point cloud is obtained as shown in formula (14):

[0124] (14)

[0125] Use the probability of formula (14) to extract the building roof point cloud from the remaining recognized point cloud. Specifically, in this embodiment, when the probability is, it is the building roof point cloud.

[0126] In specific implementation, the above process can be automatically run by computer software technology.

[0127] The above embodiments are only the preferred embodiments of the present invention, and the scope of protection of the present invention cannot be limited thereby. Any non-substantive changes and substitutions made by those skilled in the art based on the present invention belong to the scope of protection required by the present invention.

Claims

1. A method for classifying and identifying building roof and wall point clouds based on UAV technology, characterized in that: The steps include: Step 1, using UAV scanning to obtain building point cloud data; Step 2, constructing a multi-scale super-voxel fusion model based on the point cloud data obtained in step 1, and fusing the point cloud; Step 3, identifying building point clouds of different faces according to the point clouds fused by the multi-scale super voxel fusion model in step 2; Step 4, calculating the normal direction of the building point clouds of different faces identified in step 3, and identifying the wall point clouds based on the normal direction information; Step 5, after removing the wall point cloud data in step 4, the point cloud data of the remaining recognition surface is obtained, and a probability model of height, density and number of unit points is constructed for the point cloud data of the remaining recognition surface, and finally the building roof point cloud is extracted from the point cloud data of the remaining recognition surface; In step 4, the wall point cloud is identified according to the normal information of the fused supervoxel block point cloud. Specifically, when the normal direction is parallel to the xoy plane, it is determined to be a wall point cloud. The step 5 is performed as follows: After removing the wall point cloud, the remaining recognition surface point cloud is obtained , using the bounding sphere algorithm to determine the radius The number of points in each bounding sphere of , get the point cloud density of each face recognition , as shown in formula (9); （9） in, is the number of points for each identified face; According to formula (9), the normalized result is obtained as shown in formula (10): （10） At the same time, the average height of each recognition surface point cloud is calculated, as shown in formula (11): （11） According to formula (11), the normalized result is obtained as shown in formula (12): （12） The number of points for each identified face Normalize it, as shown in formula (13): （13） According to equations (10), (12) and (13), the probability of extracting the building roof point cloud is obtained, as shown in equation (14): （14） The probability of formula (14) is used to extract the building roof point cloud from the remaining recognition surface point cloud.

2. The method for classifying and identifying building roof and wall point clouds based on UAV technology as claimed in claim 1, characterized in that: The construction operation flow of the multi-scale super-voxel fusion model in step 2 is as follows: Step 2.1, constructing supervoxels on the building point cloud data in step 1 to obtain an initial supervoxel block point cloud, wherein the initial supervoxel block point cloud includes planar supervoxels and non-planar supervoxels; Step 2.2: Consider planar supervoxels and non-planar supervoxels as the basic units of the region growing algorithm, and construct a multi-scale supervoxel fusion algorithm.

3. The method for classifying and identifying building roof and wall point clouds based on UAV technology as claimed in claim 1, characterized in that: The steps of constructing the multi-scale super-voxel fusion algorithm are as follows: Step 2.21, determine the normal vector of the supervoxel block point cloud; for each point in the supervoxel block point cloud , select its nearest k neighbors, which constitute the point The local neighborhood of , and then calculate the mean of the points in the local neighborhood , as shown in formula (1): （1） According to formula (1), the covariance matrix C is constructed as shown in formula (2): （2） in, are the coordinates of the points in the neighborhood, is the average coordinate of these points; Perform eigenvalue decomposition on the covariance matrix C and obtain the eigenvalue and its corresponding eigenvector , the minimum eigenvalue The corresponding eigenvector , which is the estimated normal vector; Compute all normal vectors of points in the initial supervoxel block The average , which is the normal vector of the supervoxel point cloud; Orient the normal vector and choose a reference direction , such as the viewpoint direction or the external normal vector, check the normal vector With reference direction If the angle is greater than 90 degrees, the normal vector is reversed, as shown in formula (3): （3） Then the normal vector is normalized to make it unitary, as shown in formula (4): （4） Step 2.22: Calculate the curvature of the supervoxel point cloud. Curvature is a measure of the curvature of the supervoxel surface. The three eigenvalues ​​of the eigenvalue decomposition of formula (2) are: The curvature of each point in the supervoxel block point cloud is obtained as shown in formula (5); （5） Then calculate the curvature of all points in the supervoxel point cloud The average value of the final curvature of the supervoxel point cloud is ; Step 2.23, seed supervoxel selection: sort all supervoxel block point clouds in ascending order according to the curvature value, select the supervoxel block point cloud with the smallest curvature as the seed supervoxel, as the starting point of growth, and initialize the recognition result and the visited point set; Step 2.24, growth process: starting from the seed supervoxel, search for its nearest neighbor supervoxel in turn, and grow according to the distance threshold and normal angle threshold; if the growth condition is met, the neighboring supervoxel is added to the seed set and the recognition result, and marked as visited, so as to obtain the recognition result of the planar supervoxel; ； Step 2.25: For non-planar supervoxels , and the curvature of each non-planar supervoxel is obtained and normal vector , calculate the curvature of non-planar supervoxels Each plane supervoxel curvature is increased with area The difference between （6） Computing non-planar supervoxel normals Grow each plane supervoxel normal vector with the region The angle of （7） Simultaneously calculate the distance of each non-planar supervoxel to the nearest point of each planar supervoxel identification dataset ; According to formula (6), formula (7) and distance , we get the fusion judgment parameters of non-planar supervoxel and regional growth plane recognition supervoxel, as shown in formula (8): （8） in: is the curvature difference between the non-planar supervoxel and the neighboring planar supervoxel, is the difference in normal vector angle between the non-planar supervoxel and the neighboring planar supervoxel, is the shortest distance between a non-planar supervoxel and its neighboring planar supervoxel; Step 2.26, fusion result: when each non-planar supervoxel is transformed into a planar supervoxel recognition dataset When it is the smallest, it is fused into a super voxel block point cloud of the same surface.

4. The method for classifying and identifying building roof and wall point clouds based on UAV technology as claimed in claim 3, characterized in that: The operation of step 2.26 is as follows: when the plane supervoxel is , the plane supervoxel The neighborhood non-planar supervoxel is ; According to formula (8), calculate Neighborhood non-planar supervoxels With planar supervoxels Parameters between ; when When the minimum, non-planar supervoxel With planar supervoxels Blend together.

5. The method for classifying and identifying building roof and wall point clouds based on UAV technology as claimed in claim 1, characterized in that: When the probability , it is identified as a building roof point cloud.

6. An electronic device, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the building roof and wall point cloud classification and recognition method based on UAV technology as described in any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the building roof and wall point cloud classification and recognition method based on UAV technology as described in any one of claims 1 to 5.

Images