Building three-dimensional reconstruction method and system based on point cloud
Through recursive nearest neighbor search and plane fitting methods, noise points are filtered and deleted, the problem of noise interference in point cloud data is solved, and the accuracy of building three-dimensional reconstruction is improved.
Patent Information
- Application Number
- CN202410928072.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-11
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Point cloud data is susceptible to noise interference during the acquisition process, affecting the accuracy of the three-dimensional reconstruction of the building.
The recursive nearest neighbor search method is used to filter out noise points, and determine whether the target point is a discrete point through plane fitting. After deleting the noise points, a three-dimensional model of the building is generated.
It effectively eliminates the burrs and inaccuracies in the three-dimensional reconstruction of buildings and improves the reconstruction accuracy.
Smart Images

Figure CN118918244A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of three-dimensional reconstruction, and in particular to a point cloud-based three-dimensional reconstruction method and system for a building. Background Art
[0002] Building 3D reconstruction technology refers to the conversion of buildings in the real world into digital 3D models through various technical means. 3D models can be used as digital archives to record the structure and equipment information of buildings, providing important references for subsequent maintenance and renovation. Through 3D models, maintenance personnel can intuitively understand the structure of buildings, quickly locate and solve problems. Moreover, 3D models can record and preserve detailed information of cultural heritage and historical buildings, providing a scientific basis for restoration and protection work. Combined with virtual reality and augmented reality technologies, 3D reconstruction technology provides users with a more immersive and interactive experience. In the fields of building display, real estate sales, museum display, etc., through 3D models and virtual reality technology, users can experience the space and atmosphere of buildings in person and enhance the experience effect. However, due to various internal and external factors, point cloud data is usually interfered by noise during the acquisition process. These noises mainly include environmental noise, equipment noise and errors introduced by human operation, etc. These noises have a great impact on the accuracy of building 3D reconstruction. Summary of the invention
[0003] In order to reduce the influence of noise in building three-dimensional reconstruction, the present invention provides a method for building three-dimensional reconstruction based on point cloud, the method comprising the following steps:
[0004] S1, using a 3D laser scanner to obtain point cloud data of the building, and registering the point cloud data obtained at different positions and angles to obtain a point cloud data set;
[0005] S2, randomly select multiple points from the point cloud data set, and use the selected points as the starting point to perform a recursive nearest neighbor search. The recursive nearest neighbor search stops when the change in the distance of the nearest neighbor is greater than a threshold or all the nearest neighbor points form a ring; obtain the number of points that each selected point passes through through the recursive nearest neighbor search to the point where the recursive nearest neighbor search stops, and if the number is greater than a preset value, add all the points obtained after the recursive nearest neighbor search of the selected point to the non-noise point set;
[0006] S3, repeat step S2 multiple times, record the points that belong to the point cloud data set but do not belong to the non-noise point set as the target point; in the original point cloud data set, obtain the K nearest neighbor points of the target point, perform plane fitting on the K nearest neighbor points, and determine whether the target point is a discrete point based on the obtained plane. If so, add the target point to the discrete point set;
[0007] S4, deletes points in the discrete point set from the original point cloud dataset, and generates a three-dimensional building model based on the point cloud dataset.
[0008] Preferably, the process of the recursive nearest neighbor search is:
[0009] Search for the nearest neighbor of the current point, take the nearest neighbor as the current point, and repeat the above process until the stop condition is met;
[0010] When performing a recursive nearest neighbor search, if the nearest neighbor of the nearest neighbor of the current point returns to the current point, the next nearest neighbor of the nearest neighbor of the current point is taken as the nearest neighbor of the nearest neighbor of the current point.
[0011] Preferably, judging whether the target point is a discrete point according to the obtained plane is specifically:
[0012] The maximum distance from the point on the same side of the plane as the target point among the K nearest neighbor points to the plane is calculated. If the distance from the target point to the plane is greater than the maximum distance and / or the distance from the target point to any point on the same side of the plane as the target point is greater than the maximum distance, the target point is taken as a discrete point.
[0013] Preferably, judging whether the target point is a discrete point according to the obtained plane is specifically:
[0014] Calculate the proportion of the K nearest neighbor points that are on the same side of the plane as the target point, and obtain a distance value based on the average distance of the K nearest neighbors to the plane and the proportion. If the distance from the target point to the plane is greater than the distance value, treat the target point as a discrete point.
[0015] Preferably, the randomly selecting a plurality of points from the point cloud data set is specifically:
[0016] The area range of the point cloud data set is obtained, the area range is divided into a plurality of sub-areas of different sizes, each sub-area is randomly moved, and the point closest to the center of the sub-area is used as the selected point.
[0017] On the other hand, the present invention also provides a point cloud-based three-dimensional building reconstruction system, the system comprising the following modules:
[0018] The point cloud acquisition module is used to obtain point cloud data of the building using a 3D laser scanner, and to register the point cloud data obtained at different positions and angles to obtain a point cloud data set;
[0019] The denoising module randomly selects multiple points from the point cloud data set, and uses the selected points as the starting point for recursive nearest neighbor search. The stopping condition of the recursive nearest neighbor search is that the change in the distance of the nearest neighbor is greater than a threshold or all the nearest neighbor points form a ring; the number of points that each selected point passes through through the recursive nearest neighbor search to the stop of the recursive nearest neighbor search is obtained, and if the number is greater than a preset value, all points obtained after the recursive nearest neighbor search of the selected point are added to the non-noise point set;
[0020] Iteration module, used to repeat the denoising module multiple times, record the points that belong to the point cloud data set but not to the non-noise point set as the target point; in the original point cloud data set, get the K nearest neighbor points of the target point, perform plane fitting on the K nearest neighbor points, and judge whether the target point is a discrete point based on the obtained plane. If so, add the target point to the discrete point set;
[0021] The three-dimensional generation module is used to delete points in the discrete point set from the original point cloud data set and generate a three-dimensional model of the building based on the point cloud data set.
[0022] Preferably, the process of the recursive nearest neighbor search is:
[0023] Search for the nearest neighbor of the current point, take the nearest neighbor as the current point, and repeat the above process until the stop condition is met;
[0024] When performing a recursive nearest neighbor search, if the nearest neighbor of the nearest neighbor of the current point returns to the current point, the next nearest neighbor of the nearest neighbor of the current point is taken as the nearest neighbor of the nearest neighbor of the current point.
[0025] Preferably, judging whether the target point is a discrete point according to the obtained plane is specifically:
[0026] The maximum distance from the point on the same side of the plane as the target point among the K nearest neighbor points to the plane is calculated. If the distance from the target point to the plane is greater than the maximum distance and / or the distance from the target point to any point on the same side of the plane as the target point is greater than the maximum distance, the target point is taken as a discrete point.
[0027] Preferably, judging whether the target point is a discrete point according to the obtained plane is specifically:
[0028] Calculate the proportion of the K nearest neighbor points that are on the same side of the plane as the target point, and obtain a distance value based on the average distance of the K nearest neighbors to the plane and the proportion. If the distance from the target point to the plane is greater than the distance value, treat the target point as a discrete point.
[0029] Preferably, the randomly selecting a plurality of points from the point cloud data set is specifically:
[0030] The area range of the point cloud data set is obtained, the area range is divided into a plurality of sub-areas of different sizes, each sub-area is randomly moved, and the point closest to the center of the sub-area is used as the selected point.
[0031] In addition, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described above is implemented.
[0032] In order to solve the problem that the noise in the point cloud has a great influence on the reconstruction accuracy when using the point cloud to perform three-dimensional reconstruction of the building, the present invention adopts a recursive nearest neighbor search method to filter out the noise points, delete the noise points from the point cloud, and finally use the denoised point cloud data set to perform three-dimensional reconstruction of the building, thereby effectively eliminating the problems of burrs and inaccuracies in the three-dimensional reconstruction of the building. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Figure 1 This is a flow chart of Embodiment 1;
[0034] Figure 2 It is a schematic diagram of the point cloud of a plane and one side of the plane;
[0035] Figure 3 This is a structural diagram of the second embodiment. DETAILED DESCRIPTION
[0036] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments.
[0037] It should be noted that, in the embodiments of the present application, "at least one" refers to one or more, and "more" refers to two or more. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application. The terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without paying creative work are within the scope of protection of the present application.
[0038] Embodiment 1, the present invention provides a method for 3D reconstruction of a building based on point cloud, such as Figure 1 As shown, the method comprises the following steps:
[0039] S1, using a 3D laser scanner to obtain point cloud data of the building, and registering the point cloud data obtained at different positions and angles to obtain a point cloud data set;
[0040] When reconstructing a building in three dimensions, a three-dimensional laser scanner such as LiDAR is used to collect point cloud data of the building. For large buildings, drones equipped with laser scanners can also be used to obtain point cloud data. Since buildings are generally large, point cloud data of the building needs to be photographed from multiple angles and positions. After obtaining the point cloud data photographed at various angles and positions, the point clouds are aligned to obtain a point cloud data set.
[0041] S2, randomly select multiple points from the point cloud data set, and use the selected points as the starting point to perform a recursive nearest neighbor search. The recursive nearest neighbor search stops when the change in the distance of the nearest neighbor is greater than a threshold or all the nearest neighbor points form a ring; obtain the number of points that each selected point passes through through the recursive nearest neighbor search to the point where the recursive nearest neighbor search stops, and if the number is greater than a preset value, add all the points obtained after the recursive nearest neighbor search of the selected point to the non-noise point set;
[0042] Since the sensor, environment, etc. may increase the noise of the point cloud during the shooting process, the noise may cause the surface of the building to be uneven when it is reconstructed, and may also cause blurred edges, etc. Multiple points are randomly selected from the point cloud data set, and the random selection method includes but is not limited to: dividing the point cloud data into multiple intervals according to the x value range, y value range, and z value range, calculating the middle value of each interval, randomly selecting an middle value from all the middle values corresponding to the x value, randomly selecting an middle value from all the middle values corresponding to the y value, and randomly selecting an middle value from all the middle values corresponding to the y value. The three selected middle values form a spatial point, and the point in the point cloud data set closest to the spatial point is used as the randomly selected point; the above process is repeated continuously to obtain multiple selected points.
[0043] For example, the x value range is divided into 4 intervals, the y value range is divided into 3 intervals, and the z value range is divided into 6 intervals. The middle value of each interval is calculated. The x value range corresponds to 4 middle values x1, x2, x3, and x4, the y value range corresponds to 3 middle values y1, y2, and y3, and the z value range corresponds to 4 middle values z1, z2, z3, z4, z5, and z6. Randomly assume x2 from the x value range, y1 from the y value range, and z6 from the z value range. These (x2, y1, and z6) constitute a spatial point. The point closest to the middle point is P82 (Point82), and D82 is the randomly selected point. Repeat the above process to get the point to be selected.
[0044] In another embodiment, the randomly selecting multiple points from the point cloud data set is specifically:
[0045] The area range of the point cloud data set is obtained, the area range is divided into a plurality of sub-areas of different sizes, each sub-area is randomly moved, and the point closest to the center of the sub-area is used as the selected point.
[0046] For example, all point cloud data sets constitute a cube, that is, the cube includes all point cloud data. The cube is divided into multiple sub-intervals of different sizes, and then each sub-interval is randomly moved. Assume that sub-interval 1 moves a distance of 2 in the direction of (1,1,1), and sub-interval 2 moves a distance of 5 in the direction of (-1, 0.2, 5), and so on. The point closest to the center of the sub-area is selected as the point.
[0047] After obtaining the selected point, the selected point is used as the starting point for recursive nearest neighbor search. The recursive nearest neighbor search stops when the change in the nearest neighbor distance is greater than a threshold or when all nearest neighbor points form a ring. The process of the recursive nearest neighbor search is:
[0048] Search for the nearest neighbor of the current point, take the nearest neighbor as the current point, and repeat the above process until the stop condition is met;
[0049] For ease of understanding, an example is given below. Assuming that a selected point is P87, a nearest neighbor search is performed on P87, and the nearest neighbor of P87 is P90. The distance between P87 and P90 is 6. Next, the nearest neighbor of P90 is searched to obtain P502. The distance between P90 and P502 is 5. Next, a nearest neighbor search is performed on P502 to obtain P87. Since P87-P90-P502-P87 forms a loop, the recursive nearest neighbor search starting from P87 is stopped. Similarly, the recursion is stopped if the distance change is greater than a threshold. In a more specific embodiment, the threshold is the average value of all nearest neighbor distances in the point cloud data.
[0050] In the recursive nearest neighbor search, if the nearest neighbors of two points are each other, in order to prevent this situation, when performing a recursive nearest neighbor search, if the nearest neighbor of the nearest neighbor of the current point returns to the current point, the next nearest neighbor of the nearest neighbor of the current point is used as the nearest neighbor of the nearest neighbor of the current point. For example, the nearest neighbor of P87 is P90, and the nearest neighbor of P90 is also P87. At this time, it is necessary to find the next nearest neighbor of P90 and use the next nearest neighbor found as the nearest neighbor of P90.
[0051] Noise points are generally isolated points that are far away from other points, but there may also be two isolated components or members that are far apart but still not noise points. This requires determining the number of points after each selected point passes the recursive nearest neighbor search. If the number is greater than a preset value, all points obtained after the recursive nearest neighbor search of the selected point are added to the non-noise point set, and the remaining points may be noise points.
[0052] Still taking the above as an example, the selected point P87, after recursive nearest neighbor search, all the points obtained are P87-P90-P502, and their number is greater than 2, then these three points are deleted from the point cloud data set. It should be noted that the above data is only used to illustrate the principle of the present invention, and does not constitute a limitation of the present invention.
[0053] S3, repeat step S2 multiple times, record the points that belong to the point cloud data set but do not belong to the non-noise point set as the target point; in the original point cloud data set, obtain the K nearest neighbor points of the target point, perform plane fitting on the K nearest neighbor points, and determine whether the target point is a discrete point based on the obtained plane. If so, add the target point to the discrete point set;
[0054] Each time S2 is executed, different points will be selected, and the points obtained after recursive nearest neighbor search are also different. Each time, they are added to the non-noise point set. By executing S2 multiple times, a non-noise point set can be obtained, wherein the non-noise point set has no repeated elements, for example, a Set type set is used. In another embodiment, in order to speed up the calculation speed, before each execution of step S2, it is determined whether the non-noise point set is empty. If it is not empty, the points in the non-noise point set are removed from the point cloud data set, and the remaining points are used as the point cloud data set, and then S2 is executed.
[0055] If a point in the point cloud data set is not in the non-noise point set, it is reasonable to believe that it may be a noise point, but further judgment is required. Specifically, a point in the non-noise point set in the point cloud data set is used as a target point, and the K nearest neighbor points of the target point are obtained in the original point cloud data set, that is, the point cloud data set obtained by S1. The K nearest neighbor points are plane-fitted to obtain the plane corresponding to the K nearest neighbor points. The target point is not considered when performing plane fitting to avoid the influence of the target point as a noise point on subsequent judgment.
[0056] After obtaining the plane, determine whether the target point is a discrete point based on the plane. In one embodiment, calculate the maximum distance from the point on the same side of the plane as the target point among the K nearest neighbor points to the plane. If the distance from the target point to the plane is greater than the maximum distance and / or the distance from the target point to any point on the same side of the plane as the target point is greater than the maximum distance, then treat the target point as a discrete point.
[0057] The plane divides the space into two parts. When judging whether a target point is a discrete point, the points on the same side of the curve should be considered. The maximum distance from the points on the same side of the plane as the target point among the K nearest neighbors to the plane is calculated. For example, 5 of the K nearest neighbors are on the same side of the plane as the target point. The maximum distance from these 5 points to the plane is calculated. If the distance from the target point to the plane is greater than the maximum value, and / or the distance from the target point to any of the 5 points is greater than the maximum value, the target point is regarded as a discrete point. K is a positive integer greater than 1. There are two special cases. One is that the target point is just on the plane, then the target point is not regarded as a noise point. The other is that the number of the K nearest neighbors on the same side of the plane as the target point is less than 2, that is, 0 or 1, then the target point is directly regarded as a noise point. Figure 2 The distribution of points on the same side as the target point is shown, where the darker point is the target point, and the other points are the points on the same side of the plane as the target point among the K nearest neighbors.
[0058] In another embodiment, the proportion of points on the same side of the plane as the target point among the K nearest neighbor points is calculated, and a distance value is obtained based on the average distance of the K nearest neighbors to the plane and the proportion. If the distance from the target point to the plane is greater than the distance value, the target point is treated as a discrete point.
[0059] If most of the points on the same side of the plane as the target point among the K nearest neighbor points are target points, then the possibility that they are noise points is relatively large. Conversely, if most of the points on the same side of the plane as the target point among the K nearest neighbor points are not target points, then the possibility that they are noise points is relatively small. The proportion of points on the same side of the plane as the target point among the K nearest neighbor points that are target points can be used to determine the extent to which the K nearest neighbor points are on the same side of the plane as the target point and are target points. The distance value is obtained based on the average distance of the K nearest neighbors to the plane and the proportion. In a more specific embodiment, the distance value is calculated as (1-r)d, where r is the proportion and d is the average distance. Of course, it is not limited to this calculation method.
[0060] Among them, the proportion of the points on the same side of the plane as the target point among the K nearest neighbor points is calculated, which refers to the ratio of the number of points on the same side of the plane as the target point among the K nearest neighbor points that do not belong to the non-noise point set to the number of points on the same side of the plane as the target point among the K nearest neighbor points. For example, K = 10, there are 6 points on the same side of the plane as the target point among the 10 nearest neighbor points, and 4 of these 6 points do not belong to the non-noise point set, so the proportion is 4 / 6.
[0061] S4, deletes points in the discrete point set from the original point cloud dataset, and generates a three-dimensional building model based on the point cloud dataset.
[0062] In the original point cloud data set, that is, the point cloud data set obtained by S1, the points in the discrete point set are deleted to obtain a denoised point cloud data set, and the denoised point cloud data set is used to generate a three-dimensional model of the building.
[0063] Embodiment 2, the present invention provides a point cloud-based 3D building reconstruction system, such as Figure 3 As shown, the system includes the following modules:
[0064] The point cloud acquisition module is used to obtain point cloud data of the building using a 3D laser scanner, and to register the point cloud data obtained at different positions and angles to obtain a point cloud data set;
[0065] The denoising module randomly selects multiple points from the point cloud data set, and uses the selected points as the starting point for recursive nearest neighbor search. The stopping condition of the recursive nearest neighbor search is that the change in the distance of the nearest neighbor is greater than a threshold or all the nearest neighbor points form a ring; the number of points that each selected point passes through through the recursive nearest neighbor search to the stop of the recursive nearest neighbor search is obtained, and if the number is greater than a preset value, all points obtained after the recursive nearest neighbor search of the selected point are added to the non-noise point set;
[0066] Iteration module, used to repeat the denoising module multiple times, record the points that belong to the point cloud data set but not to the non-noise point set as the target point; in the original point cloud data set, get the K nearest neighbor points of the target point, perform plane fitting on the K nearest neighbor points, and judge whether the target point is a discrete point based on the obtained plane. If so, add the target point to the discrete point set;
[0067] The three-dimensional generation module is used to delete points in the discrete point set from the original point cloud data set and generate a three-dimensional model of the building based on the point cloud data set.
[0068] Preferably, the process of the recursive nearest neighbor search is:
[0069] Search for the nearest neighbor of the current point, take the nearest neighbor as the current point, and repeat the above process until the stop condition is met;
[0070] When performing a recursive nearest neighbor search, if the nearest neighbor of the nearest neighbor of the current point returns to the current point, the next nearest neighbor of the nearest neighbor of the current point is taken as the nearest neighbor of the nearest neighbor of the current point.
[0071] Preferably, judging whether the target point is a discrete point according to the obtained plane is specifically:
[0072] The maximum distance from the point on the same side of the plane as the target point among the K nearest neighbor points to the plane is calculated. If the distance from the target point to the plane is greater than the maximum distance and / or the distance from the target point to any point on the same side of the plane as the target point is greater than the maximum distance, the target point is taken as a discrete point.
[0073] Preferably, judging whether the target point is a discrete point according to the obtained plane is specifically:
[0074] Calculate the proportion of the K nearest neighbor points that are on the same side of the plane as the target point, and obtain a distance value based on the average distance of the K nearest neighbors to the plane and the proportion. If the distance from the target point to the plane is greater than the distance value, treat the target point as a discrete point.
[0075] Preferably, the randomly selecting a plurality of points from the point cloud data set is specifically:
[0076] The area range of the point cloud data set is obtained, the area range is divided into a plurality of sub-areas of different sizes, each sub-area is randomly moved, and the point closest to the center of the sub-area is used as the selected point.
[0077] Embodiment 3, the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method described in embodiment 1 is implemented.
[0078] It should be noted that a part of the electronic device of the above embodiment can also be implemented by a computer. In this case, a program for implementing the control function can be recorded in a computer-readable recording medium, and the program recorded in the recording medium can be read into a computer system and executed to implement it.
[0079] It should be noted that the "computer system" mentioned here refers to a computer system built into an electronic device, and uses a computer system including hardware such as an OS and peripheral devices. In addition, "computer-readable recording medium" refers to removable media such as floppy disks, magneto-optical disks, ROMs, CD-ROMs, and storage devices such as hard disks built into computer systems.
[0080] Furthermore, "computer-readable recording media" may include: media that dynamically store programs for a short period of time, such as communication lines when sending programs via networks such as the Internet or communication lines such as telephone lines; and media that store programs for a fixed period of time, such as volatile memories inside computer systems that serve as servers or clients in this case. In addition, the above-mentioned program may be a program for realizing a part of the above-mentioned functions, or a program that can realize the above-mentioned functions by combining with a program already recorded in a computer system.
[0081] In addition, the electronic device in the above-mentioned embodiment can also be implemented as a collection (device group) composed of multiple devices. Each device constituting the device group can have a part or all of the functions or functional blocks of the electronic device in the above-mentioned embodiment. As a device group, it is sufficient to have all the functions or functional blocks of the electronic device.
[0082] Those skilled in the art should recognize that the above embodiments are only used to illustrate the present application and are not intended to be limiting of the present application. As long as they are within the spirit and scope of the present application, appropriate changes and modifications to the above embodiments are within the scope of protection claimed in the present application.
Claims
1. A method for three-dimensional reconstruction of a building based on point cloud, characterized in that: The method comprises the following steps: S1, using a 3D laser scanner to obtain point cloud data of the building, and registering the point cloud data obtained at different positions and angles to obtain a point cloud data set; S2, randomly select multiple points from the point cloud data set, and use the selected points as the starting point to perform a recursive nearest neighbor search. The recursive nearest neighbor search stops when the change in the distance of the nearest neighbor is greater than a threshold or all the nearest neighbor points form a ring; obtain the number of points that each selected point passes through through the recursive nearest neighbor search to the point where the recursive nearest neighbor search stops, and if the number is greater than a preset value, add all the points obtained after the recursive nearest neighbor search of the selected point to the non-noise point set; S3, repeat step S2 multiple times, record the points that belong to the point cloud data set but do not belong to the non-noise point set as the target point; in the original point cloud data set, obtain the K nearest neighbor points of the target point, perform plane fitting on the K nearest neighbor points, and determine whether the target point is a discrete point based on the obtained plane. If so, add the target point to the discrete point set; S4, deletes points in the discrete point set from the original point cloud dataset, and generates a three-dimensional building model based on the point cloud dataset.
2. The method according to claim 1, characterized in that The process of the recursive nearest neighbor search is: Search for the nearest neighbor of the current point, take the nearest neighbor as the current point, and repeat the above process until the stop condition is met; When performing a recursive nearest neighbor search, if the nearest neighbor of the nearest neighbor of the current point returns to the current point, the next nearest neighbor of the nearest neighbor of the current point is taken as the nearest neighbor of the nearest neighbor of the current point.
3. The method according to claim 1, characterized in that The method of judging whether the target point is a discrete point according to the obtained plane is specifically as follows: The maximum distance from the point on the same side of the plane as the target point among the K nearest neighbor points to the plane is calculated. If the distance from the target point to the plane is greater than the maximum distance and / or the distance from the target point to any point on the same side of the plane as the target point is greater than the maximum distance, the target point is taken as a discrete point.
4. The method according to claim 1, characterized in that The method of judging whether the target point is a discrete point according to the obtained plane is specifically as follows: Calculate the proportion of the K nearest neighbor points that are on the same side of the plane as the target point, and obtain a distance value based on the average distance of the K nearest neighbors to the plane and the proportion. If the distance from the target point to the plane is greater than the distance value, treat the target point as a discrete point.
5. The method according to claim 1, characterized in that The method of randomly selecting multiple points from the point cloud data set is as follows: The area range of the point cloud data set is obtained, the area range is divided into a plurality of sub-areas of different sizes, each sub-area is randomly moved, and the point closest to the center of the sub-area is used as the selected point.
6. A point cloud-based 3D building reconstruction system, characterized in that: The system includes the following modules: The point cloud acquisition module is used to obtain point cloud data of the building using a 3D laser scanner, and to register the point cloud data obtained at different positions and angles to obtain a point cloud data set; The denoising module randomly selects multiple points from the point cloud data set, and uses the selected points as the starting point for recursive nearest neighbor search. The stopping condition of the recursive nearest neighbor search is that the change in the distance of the nearest neighbor is greater than a threshold or all the nearest neighbor points form a ring; the number of points that each selected point passes through through the recursive nearest neighbor search to the stop of the recursive nearest neighbor search is obtained, and if the number is greater than a preset value, all points obtained after the recursive nearest neighbor search of the selected point are added to the non-noise point set; Iteration module, used to repeat the denoising module multiple times, record the points that belong to the point cloud data set but not to the non-noise point set as the target point; in the original point cloud data set, get the K nearest neighbor points of the target point, perform plane fitting on the K nearest neighbor points, and judge whether the target point is a discrete point based on the obtained plane. If so, add the target point to the discrete point set; The three-dimensional generation module is used to delete points in the discrete point set from the original point cloud data set and generate a three-dimensional model of the building based on the point cloud data set.
7. The system according to claim 6, characterized in that The process of the recursive nearest neighbor search is: Search for the nearest neighbor of the current point, take the nearest neighbor as the current point, and repeat the above process until the stop condition is met; When performing a recursive nearest neighbor search, if the nearest neighbor of the nearest neighbor of the current point returns to the current point, the next nearest neighbor of the nearest neighbor of the current point is taken as the nearest neighbor of the nearest neighbor of the current point.
8. The system according to claim 6, characterized in that The method of judging whether the target point is a discrete point according to the obtained plane is specifically as follows: The maximum distance from the point on the same side of the plane as the target point among the K nearest neighbor points to the plane is calculated. If the distance from the target point to the plane is greater than the maximum distance and / or the distance from the target point to any point on the same side of the plane as the target point is greater than the maximum distance, the target point is taken as a discrete point.
9. The system according to claim 6, characterized in that The method of judging whether the target point is a discrete point according to the obtained plane is specifically as follows: Calculate the proportion of the K nearest neighbor points that are on the same side of the plane as the target point, and obtain a distance value based on the average distance of the K nearest neighbors to the plane and the proportion. If the distance from the target point to the plane is greater than the distance value, treat the target point as a discrete point.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the method according to any one of claims 1 to 5.
Citation Information
Cited By
Data reconstruction method and system for building house remote sensing surveying and mapping model
CN120495538A
Building live-action three-dimensional reconstruction system based on point cloud data
CN121982207A
