CN116071530B - Building roof voxelized segmentation method based on airborne laser point cloud - Google Patents
Building roof voxelized segmentation method based on airborne laser point cloud Download PDFInfo
- Publication number
- CN116071530B CN116071530B CN202310206932.0A CN202310206932A CN116071530B CN 116071530 B CN116071530 B CN 116071530B CN 202310206932 A CN202310206932 A CN 202310206932A CN 116071530 B CN116071530 B CN 116071530B
- Authority
- CN
- China
- Prior art keywords
- voxel
- points
- coordinate
- voxels
- laser
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
- G06T19/20—Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
Landscapes
- Engineering & Computer Science (AREA)
- Architecture (AREA)
- Computer Graphics (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Generation (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a building roof voxelization segmentation method based on an airborne laser point cloud. Firstly, building points in an airborne laser point cloud are extracted by using a ground filtering and classifying algorithm, all the building points are segmented into voxels, the coordinate center and normal vector characteristics of the voxels are calculated, then the building points are divided into voxel units, an initial building roof segmentation result is generated by using a region growing algorithm, the voxels are used as nodes, an energy function is built by combining the initial segmentation result and neighborhood voxel characteristic differences, and finally a final building roof segmentation result is obtained by using a graph cutting algorithm. According to the invention, the onboard laser point cloud is utilized to automatically segment the roof structure of the building, the sparse and discrete characteristics of the onboard laser point cloud are overcome, the accurate segmentation of the roof structure of the building in a complex urban scene is realized, the manual intervention can be reduced, and the digital twin, the reconstruction of the urban building model and the like can be served.
Description
Technical Field
The invention belongs to the technical field of laser scanning data processing, and particularly relates to a building roof voxelization segmentation method based on an airborne laser point cloud.
Background
The three-dimensional laser scanning can scan the earth surface in a top-down mode, can rapidly acquire laser point clouds with geometric coordinates and attributes, and can describe the real world in a digital mode in an active scanning mode, so that the three-dimensional laser scanning becomes an important three-dimensional data acquisition mode and is successfully applied to various geographic information applications. Compared with the two-dimensional image, the three-dimensional laser point cloud can acquire accurate height and geometric structure information of the urban ground object entity. The airborne three-dimensional laser scanning system can rapidly acquire three-dimensional coordinate information of a large-scale city entity, and obtains the city ground object entity structure based on the acquired three-dimensional laser point cloud.
With the advancement of digital twin technology, it is necessary to extract a city building solid model with rich details, and meanwhile, the generation of a refined city building model becomes the foundation of digital substrate construction. In order to better represent urban entities, the live-action three-dimensional model contains the three-dimensional constitutive structure of the entity and the semantic relationships at the component level.
Urban building is complex in composition, and the roofs of high-rise buildings can contain different floor structures; low-rise building roofs may contain roof structures with different slopes. For complex urban building scenes, fine extraction of different structures of the building is crucial for live-action three-dimensional reconstruction. The applications of urban solar energy analysis, disaster emergency, change detection and the like all have requirements on accurate roof structure semantic geometric relationship reconstruction results. Therefore, how to efficiently extract different structures of a building and to effectively divide a building roof in order to construct a finer building model becomes a difficult problem for research.
Disclosure of Invention
Because the point cloud has the characteristic of discreteness, the traditional point cloud clustering or segmentation algorithm is difficult to effectively extract different structures of a building, and a large amount of manual editing is needed. Aiming at the defects of the prior art, the invention provides the building roof voxelized segmentation method based on the airborne laser point cloud, which does not need manual editing, can automatically and accurately segment the building roof, and improves the reliability of automatic modeling.
In order to achieve the above purpose, the technical scheme provided by the invention is a building roof voxelization segmentation method based on an airborne laser point cloud, which comprises the following steps:
step 1, extracting building points in an airborne laser point cloud by using a ground filtering and classifying algorithm;
step 2, dividing all the building points obtained in the step 1 into voxels, and calculating the coordinate center and normal vector of the voxels;
step 3, generating an initial building roof segmentation result by using a region growing algorithm by taking voxels as units;
and 4, constructing an energy function by taking the voxels as nodes and combining the initial segmentation result and the neighborhood voxel characteristic difference, and obtaining a final building roof segmentation result by using a graph segmentation algorithm.
In the step 1, ground points are filtered out by a filtering algorithm for airborne laser point clouds to obtain non-ground points, and then the non-ground points are marked as building points and non-building points point by using a point cloud classification algorithm.
The definition of the airborne laser three-dimensional point cloud is as follows:
（1）
in the method, in the process of the invention,for the number of laser points in the airborne laser three-dimensional point cloud,/-for>Is the third point in the three-dimensional point cloudkPoint(s) of (E)>Is the third point in the three-dimensional point cloudkThe coordinate values of the corresponding space rectangular coordinate system X, Y, Z axes of the points respectively, wherein the space rectangular coordinate system takes any laser point as the origin of coordinates, the X axis points to the east, the Y axis points to the north, and the Z axis forms a right hand system along the normal direction of the ellipsoid of the earth.
In the step 2, the octree structure is used for indexing the building points, the voxels are regularly divided in the point cloud space according to the given voxel size, and the coordinate center and the normal vector of the voxels are calculated by the voxels.
Coordinates of voxelsCenter of the machineThe calculation method is as follows:
（2）
in the method, in the process of the invention,for voxel->The coordinate centers of all points in the interior are defined by voxel +.>Vector consisting of spatial coordinate average values of all laser points in the field, < >>For voxel->Total number of inner laser spots +.>For voxel->Inner firsttCoordinate values of the laser points corresponding to the X axis, < >>For voxel->Inner firsttCoordinate values of the laser points corresponding to the Y axis, < >>For voxel->Inner firsttCoordinate values corresponding to the laser points on the Z axis.
The normal vector calculation method of the voxels is as follows:
firstly, voxels are utilizedAll laser spots contained in the laser beam are used for constructing covariance matrix +.>The method comprises the following steps:
（3）
in the method, in the process of the invention,for voxel->Total number of inner laser spots +.>For voxel->Inner firsttVectors composed of spatial coordinates of the individual laser points, +.>，/>For voxel->Inner firsttCoordinate values of the laser points corresponding to the X axis, < >>For voxel->Inner firsttCoordinate values of the laser points corresponding to the Y axis, < >>For voxel->Inner firsttCoordinate values corresponding to the laser points on the Z axis, < >>For voxel->The coordinate centers of all points in the interior are defined by voxel +.>Vector consisting of spatial coordinate average values of all laser points in the field, < >>，/>For voxel->Coordinate mean value of all laser spots in X-axis, ->For voxel->Coordinate mean value of all laser spots in Y-axis, ->For voxel->The mean of the coordinates of all laser points in the Z-axis.
Covariance matrix is then formedThree corresponding eigenvalues are obtained through eigenvalue decomposition，/>Eigenvalue->The corresponding feature vector is voxel->Corresponding normal vector->。
And in the step 3, the horizontal and vertical distances between the voxel coordinate center and all the neighborhood voxel coordinate centers and the normal vector included angles between the voxel coordinate center and the neighborhood voxels are calculated respectively, a horizontal distance threshold value, a vertical distance threshold value and a normal vector included angle threshold value of the neighborhood voxel coordinate center are set, the threshold values are used as constraint conditions, the voxels are used as units, and the neighborhood voxels meeting the constraint conditions are grown by using a region growing algorithm, so that an initial building roof segmentation result is obtained.
The horizontal direction distance is calculated by the following formula:
（4）
in the method, in the process of the invention,for voxel->Coordinate center and adjacent voxels->Horizontal distance of coordinate center, +.>For voxel->Coordinate mean value of all laser spots in X-axis, ->For voxel->Coordinate mean value of all laser spots in Y-axis, ->For voxel->Coordinate mean value of all laser spots in X-axis, ->For voxel->The average of the coordinates of all laser points in the Y-axis.
The vertical distance is calculated by the following formula:
（5）
in the method, in the process of the invention,for voxel->Coordinate center and adjacent voxels->The vertical distance between the coordinate centers is +.>Voxel->Coordinate mean value of all laser spots in Z-axis, < >>For voxel->The mean of the coordinates of all laser points in the Z-axis.
Angle of normal vectorThe calculation mode of (2) is as follows:
（6）
（7）
in the method, in the process of the invention,for voxel->Corresponding normal vector, ++>For voxel->Corresponding normal vector, ++>，，/>Is the number of building voxels.
Given a horizontal distance thresholdVertical distance threshold>Normal vector angle threshold->If the condition is satisfiedAnd->And->Then voxel is made +.>And voxel->And (5) growing to generate an initial segmentation result.
Moreover, the graph segmentation algorithm in the step 4 is a segmentation method based on graph theory, the core idea is energy optimization, and the graph segmentation algorithm is widely applied to foreground and background segmentation of images in the field of computer vision. Each voxel is subjected toAs a node, optimizing an initial building roof segmentation result by using a graph segmentation algorithm, namely taking the initial segmentation result obtained in the step 3 as an initial label, constructing an energy function, taking the minimum energy function as an optimization target, and obtaining a new segmentation label of each voxel by using the graph segmentation algorithm to obtain a final building roof segmentation result; the energy function is calculated as follows, and consists of data items and smooth items:
（8）
in the method, in the process of the invention,is a label corresponding to all voxels, < >>For the number of building voxelsQuantity (S)>Representing voxel->The label of (2) is->The cost value in time, N is the set of the current voxel and all neighborhood voxel group pairs, wherein the neighborhood voxel is voxel +.>26 neighborhood voxels spatially adjacent in the direction X, Y, Z, respectively, < >>Index for adjacent voxels, < >>Is voxel->And neighborhood voxel->The labels of (2) are->And->Cost value of (a).
Data itemDefined by the initial segmentation result similarity to region growth:
（9）
in the method, in the process of the invention,for voxel->Is used for the initial tag of (a),ais voxel->Label->With the original tag->Inconsistent penalty values.
Smoothing terms generated by region growing processTo optimize the over-segmentation phenomenon due to region growth, the smoothing term is therefore defined as:
（10）
（11）
in the method, in the process of the invention,for voxel->Coordinate center and neighborhood voxel->Three-dimensional distance of coordinate center->Is a voxelAnd neighborhood voxel->Is arranged at the angle of the normal vector,bandcpenalty term for tag inconsistency, +.>For voxel->Coordinate center and neighborhood voxel->Three-dimensional distance threshold of coordinate center, +.>For voxel->And neighborhood voxel->A normal vector angle threshold.
And obtaining the optimal labels corresponding to all the voxels through minimizing the energy function, wherein the voxels with the same label finally are the same building roof structure.
Compared with the prior art, the invention has the following advantages:
1) The building roof structure is automatically segmented by utilizing the airborne laser point cloud, the point cloud is divided into voxels and segmented by taking the voxels as units, the sparse and discrete characteristics of the airborne point cloud are overcome, the building roof structure can be accurately segmented, the manual intervention can be reduced, and the efficiency of live-action three-dimensional reconstruction is improved.
2) After dividing the building point cloud into voxels, dividing the building roof structure from thick to thin by using the voxels as units and using a region growing algorithm and a graph cutting algorithm.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention.
Detailed Description
The invention provides a building roof voxelization segmentation method based on an onboard laser point cloud.
The technical scheme of the invention is further described below with reference to the accompanying drawings and examples.
As shown in fig. 1, the flow of the embodiment of the present invention includes the following steps:
and step 1, extracting building points in the airborne laser point cloud by using a ground filtering and classifying algorithm.
Firstly, filtering ground points of an airborne laser point cloud by using a cloth filtering algorithm to obtain non-ground points, and then marking the non-ground points as building points and non-building points point by using a point cloud classification algorithm.
The definition of the airborne laser three-dimensional point cloud is as follows:
（1）
in the method, in the process of the invention,for the number of laser points in the airborne laser three-dimensional point cloud,/-for>Is the third point in the three-dimensional point cloudkPoint(s) of (E)>Is the third point in the three-dimensional point cloudkThe coordinate values of the corresponding space rectangular coordinate system X, Y, Z axes of the points respectively, wherein the space rectangular coordinate system takes any laser point as the origin of coordinates, the X axis points to the east, the Y axis points to the north, and the Z axis forms a right hand system along the normal direction of the ellipsoid of the earth.
And 2, dividing all the building points obtained in the step 1 into voxels, and calculating the coordinate center and normal vector of the voxels.
And (3) establishing indexes for building points by utilizing the octree structure, dividing voxels in a point cloud space rule according to a given voxel size, and calculating the coordinate center and normal vector of the voxels by voxel.
Coordinate center of voxelThe calculation method is as follows:
（2）
in the method, in the process of the invention,for voxel->The coordinate centers of all points in the interior are defined by voxel +.>Vector consisting of spatial coordinate average values of all laser points in the field, < >>For voxel->Total number of inner laser spots +.>For voxel->Inner firsttCoordinate values of the laser points corresponding to the X axis, < >>For voxel->Inner firsttCoordinate values of the laser points corresponding to the Y axis, < >>For voxel->Inner firsttCoordinate values corresponding to the laser points on the Z axis.
The normal vector calculation method of the voxels is as follows:
firstly, voxels are utilizedAll laser spots contained in the laser beam are used for constructing covariance matrix +.>The method comprises the following steps:
（3）
in the method, in the process of the invention,for voxel->Total number of inner laser spots +.>For voxel->Inner firsttVectors composed of spatial coordinates of the individual laser points, +.>，/>For voxel->Inner firsttCoordinate values of the laser points corresponding to the X axis, < >>For voxel->Inner firsttCoordinate values of the laser points corresponding to the Y axis, < >>For voxel->Inner firsttCoordinate values corresponding to the laser points on the Z axis, < >>For voxel->The coordinate centers of all points in the interior are defined by voxel +.>Vector consisting of spatial coordinate average values of all laser points in the field, < >>，/>For voxel->Coordinate mean value of all laser spots in X-axis, ->For voxel->Coordinate mean value of all laser spots in Y-axis, ->For voxel->The mean of the coordinates of all laser points in the Z-axis.
Covariance matrix is then formedThree corresponding eigenvalues are obtained through eigenvalue decomposition，/>Eigenvalue->The corresponding feature vector is voxel->Corresponding normal vector->。
And 3, generating an initial building roof segmentation result by using the voxels as units and utilizing a region growing algorithm.
And respectively calculating the horizontal and vertical distances between the voxel coordinate center and all neighbor voxel coordinate centers and the normal vector included angles between the voxel coordinate centers and the neighbor voxels, setting a horizontal distance threshold value, a vertical distance threshold value and a normal vector included angle threshold value of the neighbor voxel coordinate centers, taking the threshold values as constraint conditions, and growing the neighbor voxels meeting the constraint conditions by using a region growing algorithm by taking the voxels as units to obtain an initial building roof segmentation result.
The horizontal direction distance is calculated by the following formula:
（4）
in the method, in the process of the invention,for voxel->Coordinate center and adjacent voxels->Horizontal distance of coordinate center, +.>For voxel->Coordinate mean value of all laser spots in X-axis, ->For voxel->Coordinate mean value of all laser spots in Y-axis, ->For voxel->Coordinate mean value of all laser spots in X-axis, ->For voxel->The average of the coordinates of all laser points in the Y-axis.
The vertical distance is calculated by the following formula:
（5）
in the method, in the process of the invention,for voxel->Coordinate center and adjacent voxels->Vertical direction of coordinate centerDistance of->Voxel->Coordinate mean value of all laser spots in Z-axis, < >>For voxel->The mean of the coordinates of all laser points in the Z-axis.
Angle of normal vectorThe calculation mode of (2) is as follows:
（6）
（7）
in the method, in the process of the invention,for voxel->Corresponding normal vector, ++>For voxel->Corresponding normal vector, ++>，，/>Is the number of building voxels.
Given a horizontal distance thresholdVertical distance threshold>Normal vector angle threshold->If the condition is satisfiedAnd->And->Then voxel is made +.>And voxel->And (5) growing to generate an initial segmentation result. In this embodiment, the horizontal distance threshold +.>Is 4m, vertical distance threshold->Is 4m, normal vector included angle threshold value +.>Is->。
And 4, constructing an energy function by taking the voxels as nodes and combining the initial segmentation result and the neighborhood voxel characteristic difference, and obtaining a final building roof segmentation result by using a graph segmentation algorithm.
And taking each voxel as a node, and optimizing the initial building roof segmentation result by using a graph segmentation algorithm. The graph cutting algorithm is a graph theory-based segmentation method, the core idea of which is energy optimization, and the graph cutting algorithm is widely applied to foreground and background segmentation of images in the field of computer vision. And (3) taking the initial segmentation result obtained in the step (3) as an initial label, constructing a data item of an energy function, constructing a smooth item of the energy function by utilizing the geometrical characteristic difference of the neighborhood voxels, taking the minimum energy function as an optimization target, and acquiring a new segmentation label of each voxel by utilizing a graph segmentation algorithm to obtain a final building roof segmentation result.
The graph cut algorithm is treated as a multi-topic problem, i.e., each voxelAll act as one node and are assigned a label in the finite set L. The optimization objective is to give each voxel +.>Assign a label->And minimizes the constructed energy function value. The energy function is calculated as follows, and consists of data items and smooth items:
（8）
in the method, in the process of the invention,is a label corresponding to all voxels, < >>For the number of building voxels +.>Representing voxel->The label of (2) is->The cost value in time, N is the set of the current voxel and all neighborhood voxel group pairs, wherein the neighborhood voxel is voxel +.>26 neighborhood voxels spatially adjacent in the direction X, Y, Z, respectively, < >>Index for adjacent voxels, < >>Is voxel->And neighborhood voxel->The labels of (2) are->And->Cost value of (a).
The data item is defined by an initial segmentation result similarity to the region growth:
（9）
in the method, in the process of the invention,for voxel->Is used for the initial tag of (a),ais voxel->Label->With the original tag->Inconsistent penalty values, this embodimentaThe value was taken as 5.
The smoothing term generated in the region growing process is to optimize the over-segmentation phenomenon caused by the region growing, so the smoothing term is defined as:
（10）
（11）
in the method, in the process of the invention,for voxel->Coordinate center and neighborhood voxel->Three-dimensional distance of coordinate center->Is a voxelAnd neighborhood voxel->Is arranged at the angle of the normal vector,bandcpenalty term for tag inconsistency, +.>For voxel->Coordinate center and neighborhood voxel->Three-dimensional distance threshold of coordinate center, +.>For voxel->And neighborhood voxel->Normal vector angle threshold, in this embodiment, three-dimensional distance threshold +.>6m, normal vector angle threshold +.>Is->Penalty valuebAndcand are all 7.
By usingThe expansion algorithm minimizes the energy function to obtain the best label corresponding to all voxels, and finally the voxels with the same label are the same building roof structure.
The specific embodiments described herein are offered by way of example only to illustrate the spirit of the invention. Those skilled in the art may make various modifications or additions to the described embodiments or substitutions thereof without departing from the spirit of the invention or exceeding the scope of the invention as defined in the accompanying claims.
Claims (5)
1. The building roof voxelization segmentation method based on the airborne laser point cloud is characterized by comprising the following steps of:
step 1, extracting building points in an airborne laser point cloud by using a ground filtering and classifying algorithm;
step 2, dividing all the building points obtained in the step 1 into voxels, and calculating the coordinate center and normal vector of the voxels;
step 3, generating an initial building roof segmentation result by using a region growing algorithm by taking voxels as units;
respectively calculating horizontal and vertical distances between the voxel coordinate center and all neighbor voxel coordinate centers and normal vector included angles between the voxel coordinate center and the neighbor voxels, setting a horizontal distance threshold value, a vertical distance threshold value and a normal vector included angle threshold value of the neighbor voxel coordinate center, taking the threshold values as constraint conditions, and growing the neighbor voxels meeting the constraint conditions by using a region growing algorithm by taking the voxels as units to obtain an initial building roof segmentation result; the horizontal direction distance is calculated by the following formula:
（4）
in the method, in the process of the invention,for voxel->Coordinate center and adjacent voxels->Horizontal distance of coordinate center, +.>Is a voxelCoordinate mean value of all laser spots in X-axis, ->For voxel->All laser points are at YThe average value of the coordinates of the axes,for voxel->Coordinate mean value of all laser spots in X-axis, ->For voxel->The average value of the coordinates of all laser points in the Y axis;
the vertical distance is calculated by the following formula:
（5）
in the method, in the process of the invention,for voxel->Coordinate center and adjacent voxels->The vertical distance between the coordinate centers is +.>Voxel->Coordinate mean value of all laser spots in Z-axis, < >>For voxel->The average value of the coordinates of all laser points in the Z axis;
step 4, using voxels as nodes, constructing an energy function by combining the initial segmentation result and the neighborhood voxel characteristic difference, and obtaining a final building roof segmentation result by using a graph segmentation algorithm;
the graph cutting algorithm is a graph theory-based segmentation method, the core idea is energy optimization, and the graph cutting algorithm is widely applied to foreground and background segmentation of images in the field of computer vision; each voxel is subjected toAs a node, optimizing an initial building roof segmentation result by using a graph segmentation algorithm, namely taking the initial segmentation result obtained in the step 3 as an initial label, constructing an energy function, taking the minimum energy function as an optimization target, and obtaining a new segmentation label of each voxel by using the graph segmentation algorithm to obtain a final building roof segmentation result; the energy function is calculated as follows, and consists of data items and smooth items:
（8）
in the method, in the process of the invention,is a label corresponding to all voxels, < >>For the number of building voxels +.>Representing voxel->The label of (2) is->The cost value at time, N, is the set of the current voxel and all neighborhood voxel group pairsWherein the neighborhood voxel is voxel->26 neighborhood voxels spatially adjacent in the direction X, Y, Z, respectively, < >>Index for adjacent voxels, < >>Is voxel->And neighborhood voxel->The labels of (2) are->And->Cost value of (2);
data itemDefined by the initial segmentation result similarity to region growth:
（9）
in the method, in the process of the invention,for voxel->Is used for the initial tag of (a),ais voxel->Label->With the original tag->Inconsistent penalty values;
smoothing terms generated by region growing processThe definition is as follows:
（10）
（11）
in the method, in the process of the invention,for voxel->Coordinate center and neighborhood voxel->Three-dimensional distance of coordinate center->For voxel->And neighborhood voxel->Is arranged at the angle of the normal vector,bandcpenalty term for tag inconsistency, +.>For voxel->Coordinate center and neighborhood voxel->Three-dimensional distance threshold of coordinate center, +.>For voxel->And neighborhood voxel->A normal vector angle threshold;
and obtaining the optimal labels corresponding to all the voxels through minimizing the energy function, wherein the voxels with the same label finally are the same building roof structure.
2. The method for voxel segmentation of building roof based on airborne laser point cloud as set forth in claim 1, wherein: in the step 1, firstly, ground points are filtered out by using a filtering algorithm for airborne laser point clouds to obtain non-ground points, then, the non-ground points are marked as building points and non-building points point by using a point cloud classification algorithm, and the definition of the airborne laser three-dimensional point clouds is as follows:
（1）
in the method, in the process of the invention,for the number of laser points in the airborne laser three-dimensional point cloud,/-for>Is the third point in the three-dimensional point cloudkPoint(s) of (E)>Is the third point in the three-dimensional point cloudkThe coordinate values of the corresponding space rectangular coordinate system X, Y, Z axes of the points respectively, wherein the space rectangular coordinate system takes any laser point as the origin of coordinates, the X axis points to the east, the Y axis points to the north, and the Z axis forms a right hand system along the normal direction of the ellipsoid of the earth.
3. The method for voxel segmentation of building roof based on airborne laser point cloud as set forth in claim 1, wherein: in step 2, building points are indexed by utilizing an octree structure, voxels are divided in a point cloud space rule according to given voxel sizes, coordinate centers and normal vectors of the voxels are calculated from voxel to voxel, and the coordinate centers of the voxels are calculatedThe calculation method is as follows:
（2）
in the method, in the process of the invention,for voxel->The coordinate centers of all points in the interior are defined by voxel +.>Vector consisting of spatial coordinate average values of all laser points in the field, < >>For voxel->Total number of inner laser spots +.>For voxel->Inner firsttCoordinate values of the laser points corresponding to the X axis, < >>For voxel->Inner firsttCoordinate values of the laser points corresponding to the Y axis, < >>For voxel->Inner firsttCoordinate values corresponding to the laser points on the Z axis.
4. A method for voxelized segmentation of building roofs based on an on-board laser point cloud as set forth in claim 3, wherein: the normal vector calculation method of the voxels in the step 2 is as follows: firstly, voxels are utilizedAll laser spots contained in the laser beam are used for constructing covariance matrix +.>The method comprises the following steps:
（3）
in the method, in the process of the invention,for voxel->Internal laser spotTotal number of (or)>For voxel->Inner firsttVectors composed of spatial coordinates of the individual laser points, +.>，/>For voxel->Inner firsttCoordinate values corresponding to the laser points on the X axis,for voxel->Inner firsttCoordinate values of the laser points corresponding to the Y axis, < >>For voxel->Inner firsttCoordinate values corresponding to the laser points on the Z axis, < >>For voxel->The coordinate centers of all points in the interior are defined by voxel +.>Vector consisting of spatial coordinate average values of all laser points in the field, < >>，/>For voxel->The average of the coordinates of all laser points in the X-axis,for voxel->Coordinate mean value of all laser spots in Y-axis, ->For voxel->The average value of the coordinates of all laser points in the Z axis;
covariance matrix is then formedThe corresponding three eigenvalues +.>，Eigenvalue->The corresponding feature vector is voxel->Corresponding normal vector->。
5. The method for voxel segmentation of building roof based on airborne laser point cloud as set forth in claim 4, wherein: normal vector included angle in step 3The calculation mode of (2) is as follows:
（6）
（7）
in the method, in the process of the invention,for voxel->Corresponding normal vector, ++>For voxel->Corresponding normal vector, ++>，，/>Number of voxels for the building;
given a horizontal distance thresholdVertical distance threshold>Normal vector angle threshold->If the condition is satisfiedAnd->And->Then voxel is made +.>And voxel->And (5) growing to generate an initial segmentation result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310206932.0A CN116071530B (en) | 2023-03-07 | 2023-03-07 | Building roof voxelized segmentation method based on airborne laser point cloud |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310206932.0A CN116071530B (en) | 2023-03-07 | 2023-03-07 | Building roof voxelized segmentation method based on airborne laser point cloud |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116071530A CN116071530A (en) | 2023-05-05 |
| CN116071530B true CN116071530B (en) | 2023-07-28 |
Family
ID=86180314
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310206932.0A Active CN116071530B (en) | 2023-03-07 | 2023-03-07 | Building roof voxelized segmentation method based on airborne laser point cloud |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116071530B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117576144B (en) * | 2024-01-15 | 2024-03-29 | 湖北工业大学 | Laser point cloud power line extraction method and device and electronic equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021097618A1 (en) * | 2019-11-18 | 2021-05-27 | 深圳市大疆创新科技有限公司 | Point cloud segmentation method and system, and computer storage medium |
| CN113205529A (en) * | 2021-04-19 | 2021-08-03 | 武汉大学 | Method for segmenting top surface of building based on airborne LiDAR point cloud |
| CN114049395A (en) * | 2021-03-31 | 2022-02-15 | 武汉大学 | Airborne point cloud assisted vehicle-mounted laser point cloud position precision improving method |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8803878B2 (en) * | 2008-03-28 | 2014-08-12 | Schlumberger Technology Corporation | Visualizing region growing in three dimensional voxel volumes |
| US8224097B2 (en) * | 2008-06-12 | 2012-07-17 | Sri International | Building segmentation for densely built urban regions using aerial LIDAR data |
| KR20130026853A (en) * | 2011-09-06 | 2013-03-14 | 한국전자통신연구원 | Apparatus and method for rendering of point cloud using voxel grid |
| CN109242855B (en) * | 2018-07-19 | 2020-08-11 | 中国科学院自动化研究所 | Multi-resolution three-dimensional statistical information-based roof segmentation method, system and equipment |
| CN109685821A (en) * | 2018-12-26 | 2019-04-26 | 中国科学院大学 | Region growing 3D rock mass point cloud plane extracting method based on high quality voxel |
| CN114463338B (en) * | 2022-01-07 | 2024-05-03 | 武汉大学 | Automatic building laser foot point extraction method based on graph cutting and post-processing |
| CN114758323A (en) * | 2022-06-13 | 2022-07-15 | 青岛市勘察测绘研究院 | Urban road sign extraction method based on vehicle-mounted laser point cloud |
| CN114764871B (en) * | 2022-06-15 | 2022-09-13 | 青岛市勘察测绘研究院 | Urban building attribute extraction method based on airborne laser point cloud |
-
2023
- 2023-03-07 CN CN202310206932.0A patent/CN116071530B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021097618A1 (en) * | 2019-11-18 | 2021-05-27 | 深圳市大疆创新科技有限公司 | Point cloud segmentation method and system, and computer storage medium |
| CN114049395A (en) * | 2021-03-31 | 2022-02-15 | 武汉大学 | Airborne point cloud assisted vehicle-mounted laser point cloud position precision improving method |
| CN113205529A (en) * | 2021-04-19 | 2021-08-03 | 武汉大学 | Method for segmenting top surface of building based on airborne LiDAR point cloud |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116071530A (en) | 2023-05-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110570428B (en) | Method and system for dividing building roof sheet from large-scale image dense matching point cloud | |
| CN111915730B (en) | Method and system for automatically generating indoor three-dimensional model by taking semantic slave point cloud into consideration | |
| CN112070769B (en) | Layered point cloud segmentation method based on DBSCAN | |
| CN109961440B (en) | Three-dimensional laser radar point cloud target segmentation method based on depth map | |
| CN107292276B (en) | Vehicle-mounted point cloud clustering method and system | |
| CN106780524B (en) | Automatic extraction method for three-dimensional point cloud road boundary | |
| Sun et al. | Aerial 3D building detection and modeling from airborne LiDAR point clouds | |
| CN111815776A (en) | Three-dimensional building fine geometric reconstruction method integrating airborne and vehicle-mounted three-dimensional laser point clouds and streetscape images | |
| CN111598916A (en) | Preparation method of indoor occupancy grid map based on RGB-D information | |
| CN105740798B (en) | A kind of point cloud object scene recognition methods based on structural analysis | |
| CN112288857A (en) | Robot semantic map object recognition method based on deep learning | |
| CN112164145B (en) | Method for rapidly extracting indoor three-dimensional line segment structure based on point cloud data | |
| CN114119902A (en) | Building extraction method based on unmanned aerial vehicle inclined three-dimensional model | |
| CN116071530B (en) | Building roof voxelized segmentation method based on airborne laser point cloud | |
| CN117132915B (en) | Power transmission line tree obstacle hidden danger analysis method based on automatic classification of point cloud | |
| CN113985435A (en) | Mapping method and system fusing multiple laser radars | |
| CN116258857A (en) | Outdoor tree-oriented laser point cloud segmentation and extraction method | |
| CN112989470B (en) | Indoor space layout extraction method based on point cloud data | |
| Guinard et al. | Piecewise-planar approximation of large 3d data as graph-structured optimization | |
| CN114463338A (en) | Automatic building laser foot point extraction method based on graph cutting and post-processing | |
| CN114463396A (en) | Point cloud registration method using plane shape and topological graph voting | |
| Sun et al. | Automated segmentation of LiDAR point clouds for building rooftop extraction | |
| Lei et al. | Automatic identification of street trees with improved RandLA-Net and accurate calculation of shading area with density-based iterative α-shape | |
| CN113536959A (en) | Dynamic obstacle detection method based on stereoscopic vision | |
| CN112633293A (en) | Three-dimensional sparse point cloud reconstruction image set classification method based on image segmentation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |