CN118072029A - Vehicle-mounted point cloud single wood segmentation method and system for improving Thiessen polygon constraint - Google Patents
Vehicle-mounted point cloud single wood segmentation method and system for improving Thiessen polygon constraint Download PDFInfo
- Publication number
- CN118072029A CN118072029A CN202410494233.5A CN202410494233A CN118072029A CN 118072029 A CN118072029 A CN 118072029A CN 202410494233 A CN202410494233 A CN 202410494233A CN 118072029 A CN118072029 A CN 118072029A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- rod
- unit
- segmentation
- concentric circle
- 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.)
- Granted
Links
- 230000011218 segmentation Effects 0.000 title claims abstract description 88
- 238000000034 method Methods 0.000 title claims abstract description 64
- 239000002023 wood Substances 0.000 title claims abstract description 41
- 238000010276 construction Methods 0.000 claims description 28
- 238000001914 filtration Methods 0.000 claims description 14
- 238000010606 normalization Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 abstract description 4
- 238000011835 investigation Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000003709 image segmentation Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000002028 Biomass Substances 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a vehicle-mounted point cloud single wood segmentation method and a vehicle-mounted point cloud single wood segmentation system for improving Thiessen polygon constraint, wherein the vehicle-mounted point cloud single wood segmentation method and the vehicle-mounted point cloud single wood segmentation system comprise the following steps of: extracting ground points and constructing a digital elevation model based on a thin plate spline interpolation method; constructing a concentric circle model by combining the characteristics of the trunk of the street tree, and positioning the position of a rod-shaped target in the three-dimensional scene by utilizing the concentric circle model and the digital elevation model; judging the types of the appendages on the positioned rod-shaped targets by using the concentric circle model, and distinguishing the pavement tree from other rod-shaped targets; and constructing a Thiessen polygon by using the trunk position of the street tree, using the Thiessen polygon area as a constraint, using the trunk as an area growth starting point, and performing single wood segmentation based on the Thiessen polygon constraint. The invention can not only greatly save labor and time cost, but also improve the accuracy and efficiency of single wood segmentation, and realize the rapid and accurate extraction and identification of the pavement trees at the two sides of the road gallery.
Description
Technical Field
The invention belongs to the field of agricultural informatization, and particularly relates to a vehicle-mounted point cloud single wood segmentation method and system for improving Thiessen polygon constraint.
Background
The forest ecosystem is a natural complex formed by a forest community and the environment thereof and having specific structure, function and self-regulation capability, and is one of the largest and most important natural ecosystems in the land ecosystem. The forest ecosystem and its changes have special ecological significance for maintaining the pattern, function and process of the natural ecosystem. The tree is used as a basic unit of a forest, and the spatial structure, biophysical and chemical components of the tree are key factors of forest accumulation and biomass estimation, species identification, tree growth modeling and the like. Thus, single wood segmentation has been one of the most critical research hotspots in forest resource investigation and management.
The traditional investigation method needs to carry out manual field investigation on the whole forest area, is time-consuming and labor-consuming, and is difficult to obtain regional and large-scale data. The vehicle-mounted laser scanning is used as a novel active remote sensing technology, can be used for measuring the horizontal and vertical structures of the stand due to the high resolution, enables the stand to be measured on a single wood level to be possible, and provides important data support for the fields of forest investigation, research, management and the like.
In recent years, the method of performing single wood segmentation by using three-dimensional point cloud data is roughly divided into two types, namely a segmentation method based on a canopy height model and a segmentation method based on three-dimensional point cloud. The segmentation method based on the canopy height model generally assumes that a tree has only one crown and the crown thereof takes on a generally conical shape. The method mainly utilizes the current image segmentation technology, such as template matching, region growing or watershed-based method, to extract and segment single trees on the rasterized canopy height model. The segmentation method based on the three-dimensional point cloud is characterized in that single wood extraction and segmentation are directly carried out on three-dimensional point cloud data, and the segmentation method based on the three-dimensional point cloud comprises a clustering method based on voxels, a K-means clustering method, a clustering method based on Markov and a global clustering method.
The prior art has the defects that: segmentation methods based on canopy-height models typically treat the single-wood segmentation problem as a gray-scale image segmentation problem, and many sophisticated image segmentation techniques can be applied to achieve superior segmentation performance. But the performance of such methods is limited by the three-dimensional information loss caused by the spatial interpolation of the three-dimensional point cloud and the rasterization process. Compared with the segmentation method based on the canopy height model, the segmentation method based on the three-dimensional point cloud is used for directly segmenting and extracting the three-dimensional point cloud, and no information is lost. Therefore, it is necessary to combine trunk position recognition and crown space structure analysis, and to directly perform single-tree segmentation in three-dimensional point cloud data.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides the vehicle-mounted point cloud single tree segmentation method and the vehicle-mounted point cloud single tree segmentation system for improving the Thiessen polygon constraint, and the vehicle-mounted point cloud single tree segmentation method and the vehicle-mounted point cloud single tree segmentation system for improving the Thiessen polygon are designed by combining trunk position identification and tree crown space structure analysis, so that not only can the labor and time cost be greatly saved, but also the accuracy and the efficiency of single tree segmentation can be improved, and the rapid and accurate extraction and identification of the pavement trees on two sides of a road gallery are realized.
In order to achieve the above object, the present invention provides the following solutions:
a vehicle-mounted point cloud single wood segmentation method for improving Thiessen polygon constraint comprises the following steps:
extracting ground points and constructing a digital elevation model based on a thin plate spline interpolation method;
Constructing a concentric circle model by combining the characteristics of the trunk of the street tree, and positioning the position of a rod-shaped target in the three-dimensional scene by utilizing the concentric circle model and the digital elevation model;
Judging the types of the appendages on the positioned rod-shaped targets by using the concentric circle model, and distinguishing the pavement tree from other rod-shaped targets;
And constructing a Thiessen polygon by using the trunk position of the street tree, using the Thiessen polygon area as a constraint, using the trunk as an area growth starting point, and performing single wood segmentation based on the Thiessen polygon constraint.
Preferably, the method for extracting ground points and constructing a digital elevation model based on a thin-plate spline interpolation method comprises the following steps:
constructing a point cloud pyramid based on the virtual grid;
filtering the point cloud pyramid;
Identifying ground points based on the filtered point cloud pyramid;
And constructing a digital elevation model based on the ground points.
Preferably, the method for locating the position of the rod-shaped object in the three-dimensional scene by using the concentric circle model comprises the following steps:
the original point cloud is subjected to difference with the generated digital elevation model, and a normalized three-dimensional point cloud is obtained;
Slicing the normalized three-dimensional point cloud at equal intervals along the Z-axis direction, and performing Euclidean clustering on the point clouds in adjacent slices to obtain a series of point clusters;
and vertically projecting each point cluster downwards along the Z axis, wherein if all points of the current point cluster are in the inner circle of the concentric circle model, the current point cluster is the position of the rod-shaped target, otherwise, the current point cluster is the position of the non-rod-shaped target.
Preferably, the method for distinguishing the street tree from other rod-shaped objects by judging the type of the appendages on the positioned rod-shaped objects by using the concentric circle model comprises the following steps:
Identifying the rod-shaped target from bottom to top by utilizing the concentric circle model;
if the rod-shaped object is a street tree, the concentric circle model presents linear characteristics at the trunk part, after the position of the crown is reached, the concentric circle model does not identify a linear structure any more, and the projection area of the nonlinear structure on the horizontal plane reaches the preset requirement;
If the rod-shaped object is other rod-shaped ground objects, the linear structure can still be identified after the concentric circle model passes through the appendages on the other rod-shaped ground objects, and the projection area of the nonlinear structure on the horizontal plane is smaller than the preset requirement.
Preferably, the method for performing single wood segmentation based on Thiessen polygonal constraint by taking a Thiessen polygonal region as a constraint and taking a trunk as a region growing starting point comprises the following steps:
automatically constructing a Delaunay triangulation network according to the trunk positions of the identified street trees;
Traversing each side of the triangle in the Delaunay triangle network, and making a section perpendicular to the XOY plane on the crown point cloud along the side of the triangle;
Connecting local maximum points on the profile to obtain a section line of the top of the crown;
calculating the position of the minimum value on the section line to be the optimal dividing point of the triangle side;
making a perpendicular line of the triangle side by using the optimal division point, and dividing the region;
And taking the trunk position as an increase starting point, taking the Thiessen polygonal area as a constraint, and carrying out area increase in each area of the segmentation to realize single wood segmentation.
The invention also provides a vehicle-mounted point cloud single wood segmentation system for improving Thiessen polygon constraint, which comprises the following steps: the device comprises a construction module, a positioning module, a classification module and a segmentation module;
the construction module is used for extracting ground points and constructing a digital elevation model based on a thin plate spline interpolation method;
the positioning module is used for constructing a concentric circle model by combining the characteristics of the trunk of the street tree, and positioning the position of a rod-shaped target in the three-dimensional scene by utilizing the concentric circle model and the digital elevation model;
The classification module is used for judging the types of the appendages on the positioned rod-shaped targets by utilizing the concentric circle model and distinguishing the pavement tree from other rod-shaped targets;
the segmentation module is used for constructing Thiessen polygons by using trunk positions of the street trees, taking Thiessen polygon areas as constraints, taking trunks as area growth starting points, and performing single wood segmentation based on the Thiessen polygon constraints.
Preferably, the construction module includes: the device comprises a first construction unit, a filtering unit, a ground point identification unit and a second construction unit;
The first construction unit is used for constructing a point cloud pyramid based on the virtual grid;
The filtering unit is used for filtering the point cloud pyramid;
the ground point identification unit is used for identifying ground points based on the filtered point cloud pyramid;
The second construction unit is used for constructing a digital elevation model based on the ground points.
Preferably, the positioning module includes: the device comprises a normalization unit, a clustering unit and a projection unit;
The normalization unit is used for performing difference between the original point cloud and the generated digital elevation model to obtain normalized three-dimensional point cloud;
The clustering unit is used for equally spacing slicing the normalized three-dimensional point cloud along the Z-axis direction and performing Euclidean clustering on the point clouds in adjacent slices to obtain a series of point clusters;
The projection unit is used for vertically projecting each point cluster downwards along the Z axis, if all points of the current point cluster are in the inner circle of the concentric circle model, the current point cluster is the position of the rod-shaped target, otherwise, the current point cluster is the position of the non-rod-shaped target.
Preferably, the classification module includes: a target recognition unit, a first determination unit, and a second determination unit;
The target identification unit is used for identifying the rod-shaped target from bottom to top by utilizing the concentric circle model;
The first judging unit is used for displaying linear characteristics of the concentric circle model on the trunk part if the rod-shaped object is a street tree, and the concentric circle model does not identify a linear structure after reaching the crown position, and the projection area of the nonlinear structure on the horizontal plane reaches the preset requirement;
And the second judging unit is used for still identifying the linear structure after the concentric circle model passes through the appendages on other rod-shaped ground objects if the rod-shaped object is other rod-shaped ground objects, and the projection area of the nonlinear structure on the horizontal plane is smaller than the preset requirement.
Preferably, the dividing module includes: the device comprises a triangle net construction unit, a section unit, a connection unit, a calculation unit, a region segmentation unit and a single wood segmentation unit;
The triangular net construction unit is used for automatically constructing a Delaunay triangular net according to the trunk position of the identified street tree;
the profile unit is used for traversing each side of the triangle in the Delaunay triangle network, and making a profile perpendicular to the XOY plane on the crown point cloud along the side of the triangle;
the connecting unit is used for connecting local maximum points on the profile to obtain a section line of the top of the crown;
The calculating unit is used for calculating the position of the minimum value on the profile line to be the optimal dividing point of the triangle side;
The region segmentation unit is used for making a perpendicular line of the triangle side by using the optimal segmentation point to segment the region;
The single-tree segmentation unit is used for achieving single-tree segmentation by taking the trunk position as a growth starting point and taking the Thiessen polygonal area as a constraint and carrying out area growth in each segmented area.
Compared with the prior art, the invention has the beneficial effects that:
The invention can not only greatly save labor and time cost, but also improve the accuracy and efficiency of single wood segmentation, and realize the rapid and accurate extraction and identification of the pavement trees at the two sides of the road gallery. The research result of the invention provides a beneficial reference for the application of the laser radar in single wood identification and classification, and enables the fields of forestry resource investigation, forest resource monitoring and the like.
Drawings
In order to more clearly illustrate the technical solutions of the present invention, the drawings that are needed in the embodiments are briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a digital elevation model construction based on thin-plate spline interpolation in accordance with an embodiment of the present invention;
FIG. 2 is a schematic diagram of a concentric circular model positioning rod-like object in accordance with an embodiment of the present invention;
FIG. 3 is a diagram illustrating the differentiation of a pavement tree from other rod-shaped ground objects in accordance with an embodiment of the present invention;
FIG. 4 is a diagram of single-wood segmentation based on improved Thiessen polygonal constraints, in accordance with an embodiment of the present invention;
FIG. 5 is a side view of the segmentation result of an embodiment of the present invention;
FIG. 6 is a top view of the segmentation result of an embodiment of the present invention;
Fig. 7 is a schematic flow chart of a vehicle-mounted point cloud single-tree segmentation method for improving the Thiessen polygon constraint according to the embodiment of the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
Example 1
As shown in fig. 7, the invention provides a vehicle-mounted point cloud single wood segmentation method for improving the constraint of a Thiessen polygon, which comprises the following steps:
Extracting ground points and constructing a digital elevation model based on a thin plate spline interpolation method, and calculating the height of each point in the three-dimensional point cloud from the ground for the follow-up;
Constructing a concentric circle model by combining the characteristics of the trunk of the street tree, and positioning the position of a rod-shaped target in the three-dimensional scene by utilizing the concentric circle model and the digital elevation model;
Judging the types of the appendages on the positioned rod-shaped targets by using the concentric circle model, and distinguishing the street trees from other rod-shaped targets (such as street lamps and the like);
And constructing a Thiessen polygon by using the trunk position of the street tree, using the Thiessen polygon area as a constraint, using the trunk as an area growth starting point, and performing single wood segmentation based on the Thiessen polygon constraint.
In this embodiment, the method for extracting ground points and constructing a digital elevation model based on a thin-plate spline interpolation method includes:
constructing a point cloud pyramid based on the virtual grid;
filtering the point cloud pyramid;
Identifying ground points based on the filtered point cloud pyramid;
And constructing a digital elevation model based on the ground points.
In particular, woodland surfaces are often complex irregular surface structures, which presents a significant challenge in the automated separation of ground points to generate high-precision digital elevation models (Digital Elevation Model, DEM). The thin-plate spline (THIN PLATE SPLINE, TPS) is an interpolation method, which finds a smooth curved surface with minimum bending through all control points, well reflects the physical characteristics of abnormal elevation change, and can obtain ideal interpolation results when complex local deformation is constructed. In addition, the multi-scale analysis method ensures that the points Cheng Tubian are filtered layer by layer from a sparse surface, the real ground surface is gradually approximated, and the influence of the terrain gradient is reduced. Therefore, the invention adopts the earth surface curvature filtering algorithm based on multiple scales to generate the high-precision DEM. The specific process is as follows:
And constructing a point cloud pyramid based on the virtual grid. If the control points in each grid in the top-layer maximum-scale virtual grid of the point cloud pyramid are reliable ground points, the distance between the top-layer virtual grids must be larger than the maximum building size in the target area, so that the requirements can be met. The construction of the point cloud pyramid starts from the bottommost virtual grid and is generated step by step from bottom to top until the grid spacing is larger than the maximum building size in the area (the maximum grid is set to be 16 m). Each layer of grids is downsampled, 4 adjacent grids are combined into one grid, and the lowest point in the 4 grids is taken as the control point of the next-stage grid.
The filtering process based on the point cloud pyramid is just opposite to the construction process of the point cloud pyramid, control points in a virtual grid of the top layer of the pyramid are used as ground reference points, and the original ground surface is approximated step by step through a thin plate spline function (THIN PLATE SPLINE, TPS) interpolation method from top to bottom. In consideration of the specificity of local topography, TPS interpolation is carried out point by point on each stage of grids by adopting an analysis mode of a fixed window to the whole area, and the elevation of a grid control point is interpolated by fixedly taking the grid control point in the 3x3 neighborhood of the upper grid corresponding to the grid control point each time. The estimate of the thin-plate spline function is calculated by equation (1):
（1）
Wherein a, b, c are undetermined coefficients, Is a weight coefficient determined by the distance between the grid control points,，/>Representing the number of grid control points,/>Is a matter of distance/>Is generally selected by. This function is inversely proportional to distance and tends to zero away from the data point.
For the grid control points on the top layer of the point cloud pyramid, because the control points of the grid on the layer are all considered to be ground points when the point cloud pyramid is constructed, interpolation of the grid control points on the top layer is not needed, and the control points in the grid on the layer are mainly used for interpolating the grid control points on the next layer.
And (3) performing TPS interpolation on the estimated curved surface of a certain layer of grids by using the upper layer of grid control points, and calculating fitting residual errors of the control points in each grid of the current layer. When the fitting residual error of a certain control point exceeds a threshold value, replacing the point elevation value with a fitting elevation; the grid control points with the fit residuals not exceeding the limits still retain the original elevation values, and the grid control points are endowed with the ground point attributes.
And (5) identifying the ground points. Interpolation is only performed on grid control points, and LiDAR point cloud ground points and non-ground points are determined on the bottommost grid. I.e. if the absolute value of the difference between the elevation value of the current point and the elevation value of the grid is less than or equal to a given threshold valueThen this point is the ground point (denoted 0) and otherwise is the non-ground point (denoted 1).
（2）
Wherein,Representing the/>, of the current pointCoordinates/>Representing interpolated grid elevation values,/>Representing the category of the current point (i.e., ground point or non-ground point).
And (5) generating a digital elevation model. A digital elevation model is generated using the identified ground points by interpolation using thin-plate spline functions (THIN PLATE SPLINE, TPS). As shown in fig. 1.
In this embodiment, the method for locating the position of the rod-shaped object in the three-dimensional scene by using the concentric circle model includes:
The original point cloud is subjected to difference with the generated digital elevation model, so that the absolute height of each point from the ground is obtained, and a normalized three-dimensional point cloud is generated, wherein the normalized three-dimensional point cloud is shown in a formula (3);
（3）
wherein, Representing the absolute height of the current point from the ground,/>Representing the/>, of the current pointCoordinates/>Representing the local ground elevation of the current point.
Slicing the normalized three-dimensional point cloud at equal intervals along the Z-axis direction, and performing Euclidean clustering on the point clouds in adjacent slices to obtain a series of point clusters;
and vertically projecting each point cluster downwards along the Z axis, wherein if all points of the current point cluster are in the inner circle of the concentric circle model, the current point cluster is the position of the rod-shaped target, otherwise, the current point cluster is the position of the non-rod-shaped target.
As shown in fig. 2.
In this embodiment, the method for distinguishing the street tree from other rod-shaped objects by using the concentric circle model to determine the type of the appendages on the positioned rod-shaped objects includes:
Identifying the rod-shaped target from bottom to top by utilizing the concentric circle model;
if the rod-shaped object is a street tree, the concentric circle model presents linear characteristics at the trunk part, after the position of the crown is reached, the concentric circle model does not identify a linear structure any more, and the projection area of the nonlinear structure on the horizontal plane reaches the preset requirement;
If the rod-shaped object is other rod-shaped ground objects, the linear structure can still be identified after the concentric circle model passes through the appendages on the other rod-shaped ground objects, and the projection area of the nonlinear structure on the horizontal plane is smaller than the preset requirement.
In particular, in addition to the pavement tree, other rod-shaped ground objects (e.g., street lamps, etc.) have similar rod-shaped structures, which are also identified by the concentric circle model. It is noted that the topological relationship between the nonlinear appendages (e.g., crowns) of the pavement tree and its rod-like structure is different from the topological relationship between the nonlinear appendages (e.g., signboards) of other rod-like ground objects and its rod-like structure. Therefore, the invention judges the type of the accessory on the rod-shaped object by using the concentric circle model, and distinguishes the tree from other rod-shaped objects (such as street lamps and the like). For street trees, the use of concentric circle models will exhibit linear characteristics in the trunk portion from bottom to top, and when the crown position is reached, the concentric circle models will no longer identify linear structures, and the projected area of the nonlinear structures (i.e., crowns) on the horizontal plane is generally larger. Unlike the street tree, the concentric circle model can still identify the linear structure after passing through the appendages on other rod-shaped ground objects, and the projection area of the nonlinear structure on the horizontal plane is generally smaller. As shown in fig. 3.
In the embodiment, in the three-dimensional scene, adhesion conditions may exist between crown point clouds of adjacent trees, so that the under-segmentation problem occurs in the region growing method. Therefore, the invention designs a region growing method based on improved Thiessen polygon constraint, which realizes single wood segmentation. The method for performing single wood segmentation based on Thiessen polygonal constraint by taking a Thiessen polygonal region as a constraint and taking a trunk as a region growing starting point comprises the following steps:
automatically constructing a Delaunay triangulation network according to the trunk positions of the identified street trees;
Traversing each side of the triangle in the Delaunay triangle network, and making a section perpendicular to the XOY plane on the crown point cloud along the side of the triangle;
Connecting local maximum points on the profile to obtain a section line of the top of the crown;
calculating the position of the minimum value on the section line to be the optimal dividing point of the triangle side;
making a perpendicular line of the triangle side by using the optimal division point, and dividing the region;
and taking the trunk position as an increase starting point, taking the Thiessen polygonal area as a constraint, and carrying out area increase in each area of the segmentation to realize single wood segmentation. As shown in fig. 4,5 and 6.
Example two
The invention also provides a vehicle-mounted point cloud single wood segmentation system for improving Thiessen polygon constraint, which comprises the following steps: the device comprises a construction module, a positioning module, a classification module and a segmentation module;
the construction module is used for extracting ground points and constructing a digital elevation model based on a thin plate spline interpolation method;
The positioning module is used for constructing a concentric circle model by combining the characteristics of the trunk of the street tree, and positioning the position of a rod-shaped target in the three-dimensional scene by utilizing the concentric circle model and the digital elevation model;
The classification module is used for judging the types of the appendages on the positioned rod-shaped targets by utilizing the concentric circle model and distinguishing the pavement tree from other rod-shaped targets;
The segmentation module is used for constructing Thiessen polygons by using trunk positions of the street trees, taking Thiessen polygon areas as constraints, taking trunks as area growth starting points, and performing single wood segmentation based on the Thiessen polygon constraints.
In this embodiment, the building block includes: the device comprises a first construction unit, a filtering unit, a ground point identification unit and a second construction unit;
the first construction unit is used for constructing a point cloud pyramid based on the virtual grid;
the filtering unit is used for filtering the point cloud pyramid;
the ground point identification unit is used for identifying ground points based on the filtered point cloud pyramid;
the second construction unit is used for constructing a digital elevation model based on the ground points.
In this embodiment, the positioning module includes: the device comprises a normalization unit, a clustering unit and a projection unit;
the normalization unit is used for performing difference between the original point cloud and the generated digital elevation model to obtain normalized three-dimensional point cloud;
the clustering unit is used for equally spacing slicing the normalized three-dimensional point cloud along the Z-axis direction, and performing Euclidean clustering on the point clouds in adjacent slices to obtain a series of point clusters;
the projection unit is used for vertically projecting each point cluster downwards along the Z axis, if all points of the current point cluster are in the inner circle of the concentric circle model, the current point cluster is the position of the rod-shaped target, otherwise, the current point cluster is the position of the non-rod-shaped target.
In this embodiment, the classification module includes: a target recognition unit, a first determination unit, and a second determination unit;
The target identification unit is used for identifying the rod-shaped target from bottom to top by utilizing the concentric circle model;
the first judging unit is used for displaying linear characteristics of the concentric circle model on the trunk part if the rod-shaped object is a street tree, and the concentric circle model does not recognize the linear structure after reaching the crown position, and the projection area of the nonlinear structure on the horizontal plane reaches the preset requirement;
The second judging unit is used for identifying the linear structure after the concentric circle model passes through the appendages on other rod-shaped ground objects if the rod-shaped object is other rod-shaped ground objects, and the projection area of the nonlinear structure on the horizontal plane is smaller than the preset requirement.
In this embodiment, the dividing module includes: the device comprises a triangle net construction unit, a section unit, a connection unit, a calculation unit, a region segmentation unit and a single wood segmentation unit;
the triangular net construction unit is used for automatically constructing a Delaunay triangular net according to the trunk position of the identified street tree;
the profile unit is used for traversing each side of the triangle in the Delaunay triangle network and making a profile perpendicular to the XOY plane on the crown point cloud along the side of the triangle;
The connecting unit is used for connecting local maximum points on the profile to obtain a section line of the top of the crown;
the calculating unit is used for calculating the position of the minimum value on the profile line to be the optimal dividing point of the triangle side;
the region segmentation unit is used for making a perpendicular line of the triangle side by using the optimal segmentation point to segment the region;
The single-tree segmentation unit is used for realizing single-tree segmentation by taking the trunk position as a growth starting point and taking the Thiessen polygonal area as a constraint and carrying out area growth in each segmented area.
The above embodiments are merely illustrative of the preferred embodiments of the present invention, and the scope of the present invention is not limited thereto, but various modifications and improvements made by those skilled in the art to which the present invention pertains are made without departing from the spirit of the present invention, and all modifications and improvements fall within the scope of the present invention as defined in the appended claims.
Claims (10)
1. The vehicle-mounted point cloud single wood segmentation method for improving Thiessen polygonal constraint is characterized by comprising the following steps of:
extracting ground points and constructing a digital elevation model based on a thin plate spline interpolation method;
Constructing a concentric circle model by combining the characteristics of the trunk of the street tree, and positioning the position of a rod-shaped target in the three-dimensional scene by utilizing the concentric circle model and the digital elevation model;
Judging the types of the appendages on the positioned rod-shaped targets by using the concentric circle model, and distinguishing the pavement tree from other rod-shaped targets;
And constructing a Thiessen polygon by using the trunk position of the street tree, using the Thiessen polygon area as a constraint, using the trunk as an area growth starting point, and performing single wood segmentation based on the Thiessen polygon constraint.
2. The method for vehicle-mounted point cloud single-tree segmentation for improving Thiessen polygonal constraints according to claim 1, wherein the method for extracting ground points and constructing a digital elevation model based on a thin-plate spline interpolation method comprises:
constructing a point cloud pyramid based on the virtual grid;
filtering the point cloud pyramid;
Identifying ground points based on the filtered point cloud pyramid;
And constructing a digital elevation model based on the ground points.
3. The method for vehicle-mounted point cloud mono segmentation for improving Thiessen polygonal constraints according to claim 1, wherein the method for locating the position of a rod-like object in a three-dimensional scene using the concentric circle model comprises:
the original point cloud is subjected to difference with the generated digital elevation model, and a normalized three-dimensional point cloud is obtained;
Slicing the normalized three-dimensional point cloud at equal intervals along the Z-axis direction, and performing Euclidean clustering on the point clouds in adjacent slices to obtain a series of point clusters;
and vertically projecting each point cluster downwards along the Z axis, wherein if all points of the current point cluster are in the inner circle of the concentric circle model, the current point cluster is the position of the rod-shaped target, otherwise, the current point cluster is the position of the non-rod-shaped target.
4. The method for vehicle-mounted point cloud single wood segmentation for improving Thiessen polygonal constraints according to claim 1, wherein the method for distinguishing a pavement tree from other rod-shaped objects by judging the type of an attachment on the positioned rod-shaped object using the concentric circle model comprises:
Identifying the rod-shaped target from bottom to top by utilizing the concentric circle model;
if the rod-shaped object is a street tree, the concentric circle model presents linear characteristics at the trunk part, after the position of the crown is reached, the concentric circle model does not identify a linear structure any more, and the projection area of the nonlinear structure on the horizontal plane reaches the preset requirement;
If the rod-shaped object is other rod-shaped ground objects, the linear structure can still be identified after the concentric circle model passes through the appendages on the other rod-shaped ground objects, and the projection area of the nonlinear structure on the horizontal plane is smaller than the preset requirement.
5. The method for vehicle-mounted point cloud single wood segmentation for improving Thiessen polygonal constraints according to claim 1, wherein the method for single wood segmentation based on Thiessen polygonal constraints with the Thiessen polygonal region as a constraint and the trunk as a region growth starting point comprises:
automatically constructing a Delaunay triangulation network according to the trunk positions of the identified street trees;
Traversing each side of the triangle in the Delaunay triangle network, and making a section perpendicular to the XOY plane on the crown point cloud along the side of the triangle;
Connecting local maximum points on the profile to obtain a section line of the top of the crown;
calculating the position of the minimum value on the section line to be the optimal dividing point of the triangle side;
making a perpendicular line of the triangle side by using the optimal division point, and dividing the region;
And taking the trunk position as an increase starting point, taking the Thiessen polygonal area as a constraint, and carrying out area increase in each area of the segmentation to realize single wood segmentation.
6. An on-vehicle point cloud single wood segmentation system of improvement Thiessen polygon constraint, characterized in that includes: the device comprises a construction module, a positioning module, a classification module and a segmentation module;
the construction module is used for extracting ground points and constructing a digital elevation model based on a thin plate spline interpolation method;
the positioning module is used for constructing a concentric circle model by combining the characteristics of the trunk of the street tree, and positioning the position of a rod-shaped target in the three-dimensional scene by utilizing the concentric circle model and the digital elevation model;
The classification module is used for judging the types of the appendages on the positioned rod-shaped targets by utilizing the concentric circle model and distinguishing the pavement tree from other rod-shaped targets;
the segmentation module is used for constructing Thiessen polygons by using trunk positions of the street trees, taking Thiessen polygon areas as constraints, taking trunks as area growth starting points, and performing single wood segmentation based on the Thiessen polygon constraints.
7. The vehicle-mounted point cloud mono segmentation system that improves Thiessen polygonal constraints according to claim 6, wherein the build module comprises: the device comprises a first construction unit, a filtering unit, a ground point identification unit and a second construction unit;
The first construction unit is used for constructing a point cloud pyramid based on the virtual grid;
The filtering unit is used for filtering the point cloud pyramid;
the ground point identification unit is used for identifying ground points based on the filtered point cloud pyramid;
The second construction unit is used for constructing a digital elevation model based on the ground points.
8. The vehicle-mounted point cloud mono segmentation system that improves Thiessen polygonal constraints of claim 6, wherein the positioning module comprises: the device comprises a normalization unit, a clustering unit and a projection unit;
The normalization unit is used for performing difference between the original point cloud and the generated digital elevation model to obtain normalized three-dimensional point cloud;
The clustering unit is used for equally spacing slicing the normalized three-dimensional point cloud along the Z-axis direction and performing Euclidean clustering on the point clouds in adjacent slices to obtain a series of point clusters;
The projection unit is used for vertically projecting each point cluster downwards along the Z axis, if all points of the current point cluster are in the inner circle of the concentric circle model, the current point cluster is the position of the rod-shaped target, otherwise, the current point cluster is the position of the non-rod-shaped target.
9. The vehicle-mounted point cloud mono segmentation system that improves Thiessen polygonal constraints of claim 6, wherein the classification module comprises: a target recognition unit, a first determination unit, and a second determination unit;
The target identification unit is used for identifying the rod-shaped target from bottom to top by utilizing the concentric circle model;
The first judging unit is used for displaying linear characteristics of the concentric circle model on the trunk part if the rod-shaped object is a street tree, and the concentric circle model does not identify a linear structure after reaching the crown position, and the projection area of the nonlinear structure on the horizontal plane reaches the preset requirement;
And the second judging unit is used for still identifying the linear structure after the concentric circle model passes through the appendages on other rod-shaped ground objects if the rod-shaped object is other rod-shaped ground objects, and the projection area of the nonlinear structure on the horizontal plane is smaller than the preset requirement.
10. The vehicle-mounted point cloud mono segmentation system that improves Thiessen polygonal constraints of claim 6, wherein the segmentation module comprises: the device comprises a triangle net construction unit, a section unit, a connection unit, a calculation unit, a region segmentation unit and a single wood segmentation unit;
The triangular net construction unit is used for automatically constructing a Delaunay triangular net according to the trunk position of the identified street tree;
the profile unit is used for traversing each side of the triangle in the Delaunay triangle network, and making a profile perpendicular to the XOY plane on the crown point cloud along the side of the triangle;
the connecting unit is used for connecting local maximum points on the profile to obtain a section line of the top of the crown;
The calculating unit is used for calculating the position of the minimum value on the profile line to be the optimal dividing point of the triangle side;
The region segmentation unit is used for making a perpendicular line of the triangle side by using the optimal segmentation point to segment the region;
The single-tree segmentation unit is used for achieving single-tree segmentation by taking the trunk position as a growth starting point and taking the Thiessen polygonal area as a constraint and carrying out area growth in each segmented area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410494233.5A CN118072029B (en) | 2024-04-24 | 2024-04-24 | Vehicle-mounted point cloud single wood segmentation method and system for improving Thiessen polygon constraint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410494233.5A CN118072029B (en) | 2024-04-24 | 2024-04-24 | Vehicle-mounted point cloud single wood segmentation method and system for improving Thiessen polygon constraint |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118072029A true CN118072029A (en) | 2024-05-24 |
| CN118072029B CN118072029B (en) | 2024-07-19 |
Family
ID=91106191
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410494233.5A Active CN118072029B (en) | 2024-04-24 | 2024-04-24 | Vehicle-mounted point cloud single wood segmentation method and system for improving Thiessen polygon constraint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118072029B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108198190A (en) * | 2017-12-28 | 2018-06-22 | 北京数字绿土科技有限公司 | A kind of single wooden dividing method and device based on point cloud data |
| CN108650626A (en) * | 2018-05-18 | 2018-10-12 | 华南师范大学 | A kind of fingerprinting localization algorithm based on Thiessen polygon |
| CN110223314A (en) * | 2019-06-06 | 2019-09-10 | 电子科技大学 | A kind of single wooden dividing method based on the distribution of tree crown three-dimensional point cloud |
| CN112669333A (en) * | 2021-01-11 | 2021-04-16 | 四川测绘地理信息局测绘技术服务中心 | Single tree information extraction method |
| CN113313081A (en) * | 2021-07-27 | 2021-08-27 | 武汉市测绘研究院 | Road traffic rod object classification method integrating vehicle-mounted three-dimensional laser point cloud and image |
| US20220198749A1 (en) * | 2019-04-19 | 2022-06-23 | Seoul National University R&Db Foundation | System and method for monitoring forest gap using lidar survey data |
| CN115063555A (en) * | 2022-07-12 | 2022-09-16 | 湖南科技大学 | Method for extracting vehicle-mounted LiDAR point cloud street tree growing in Gaussian distribution area |
| CN115690513A (en) * | 2022-11-15 | 2023-02-03 | 江苏久智环境科技服务有限公司 | Urban street tree species identification method based on deep learning |
| CN117635936A (en) * | 2023-11-21 | 2024-03-01 | 安徽开源路桥有限责任公司 | Street tree extraction and segmentation algorithm based on vehicle-mounted laser point cloud data |
| CN117893924A (en) * | 2023-11-09 | 2024-04-16 | 电子科技大学 | Unmanned aerial vehicle laser radar point cloud single wood segmentation method based on tree crown shape |
-
2024
- 2024-04-24 CN CN202410494233.5A patent/CN118072029B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108198190A (en) * | 2017-12-28 | 2018-06-22 | 北京数字绿土科技有限公司 | A kind of single wooden dividing method and device based on point cloud data |
| CN108650626A (en) * | 2018-05-18 | 2018-10-12 | 华南师范大学 | A kind of fingerprinting localization algorithm based on Thiessen polygon |
| US20220198749A1 (en) * | 2019-04-19 | 2022-06-23 | Seoul National University R&Db Foundation | System and method for monitoring forest gap using lidar survey data |
| CN110223314A (en) * | 2019-06-06 | 2019-09-10 | 电子科技大学 | A kind of single wooden dividing method based on the distribution of tree crown three-dimensional point cloud |
| CN112669333A (en) * | 2021-01-11 | 2021-04-16 | 四川测绘地理信息局测绘技术服务中心 | Single tree information extraction method |
| CN113313081A (en) * | 2021-07-27 | 2021-08-27 | 武汉市测绘研究院 | Road traffic rod object classification method integrating vehicle-mounted three-dimensional laser point cloud and image |
| CN115063555A (en) * | 2022-07-12 | 2022-09-16 | 湖南科技大学 | Method for extracting vehicle-mounted LiDAR point cloud street tree growing in Gaussian distribution area |
| CN115690513A (en) * | 2022-11-15 | 2023-02-03 | 江苏久智环境科技服务有限公司 | Urban street tree species identification method based on deep learning |
| CN117893924A (en) * | 2023-11-09 | 2024-04-16 | 电子科技大学 | Unmanned aerial vehicle laser radar point cloud single wood segmentation method based on tree crown shape |
| CN117635936A (en) * | 2023-11-21 | 2024-03-01 | 安徽开源路桥有限责任公司 | Street tree extraction and segmentation algorithm based on vehicle-mounted laser point cloud data |
Non-Patent Citations (2)
| Title |
|---|
| 朱岩彬等: "一种层次化的车载激光点云中杆状地物提取方法研究", 《地理信息世界》, 25 December 2019 (2019-12-25) * |
| 王濮;邢艳秋;王成;习晓环;: "一种基于图割的机载LiDAR单木识别方法", 中国科学院大学学报, no. 03, 15 May 2019 (2019-05-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118072029B (en) | 2024-07-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110781827B (en) | Road edge detection system and method based on laser radar and fan-shaped space division | |
| CN110570428B (en) | Method and system for dividing building roof sheet from large-scale image dense matching point cloud | |
| CN108062517B (en) | Vehicle-mounted laser point cloud-based method for automatically extracting boundary line of unstructured road | |
| CN106529469B (en) | Unmanned aerial vehicle-mounted LiDAR point cloud filtering method based on self-adaptive gradient | |
| CN110119438B (en) | Airborne LiDAR point cloud filtering method based on active learning | |
| CN110490888B (en) | Highway geometric feature vectorization extraction method based on airborne laser point cloud | |
| CN111598823A (en) | Multi-source mobile measurement point cloud data air-ground integrated fusion method and storage medium | |
| CN111340723B (en) | Terrain-adaptive airborne LiDAR point cloud regularization thin plate spline interpolation filtering method | |
| CN111354083B (en) | Progressive building extraction method based on original laser point cloud | |
| CN111598780B (en) | Terrain adaptive interpolation filtering method suitable for airborne LiDAR point cloud | |
| CN110794413A (en) | Method and system for detecting power line of point cloud data of laser radar segmented by linear voxels | |
| CN113409332B (en) | Building plane segmentation method based on three-dimensional point cloud | |
| CN115937151B (en) | Method for judging curling degree of crop leaves | |
| CN115063555A (en) | Method for extracting vehicle-mounted LiDAR point cloud street tree growing in Gaussian distribution area | |
| KR101549155B1 (en) | Method of automatic extraction of building boundary from lidar data | |
| CN111950589B (en) | Point cloud region growing optimization segmentation method combined with K-means clustering | |
| CN114463338B (en) | Automatic building laser foot point extraction method based on graph cutting and post-processing | |
| CN116704333A (en) | Single tree detection method based on laser point cloud data | |
| CN113721254B (en) | Vehicle positioning method based on road fingerprint space association matrix | |
| CN114863064A (en) | Method and system for constructing automobile contour curved surface model | |
| CN111861946B (en) | Adaptive multi-scale vehicle-mounted laser radar dense point cloud data filtering method | |
| CN118072029B (en) | Vehicle-mounted point cloud single wood segmentation method and system for improving Thiessen polygon constraint | |
| CN117173424A (en) | Point cloud slope surface edge line identification method, system and readable storage medium | |
| Qin et al. | A voxel-based filtering algorithm for mobile LiDAR data | |
| Wang et al. | An individual tree segmentation method from mobile mapping point clouds based on improved 3-D morphological analysis |
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 |