CN106204724A - A kind of 3 D laser scanning measures the method for single wood apparatus derivatorius - Google Patents

Abstract

The invention discloses a method for measuring the branch structure of a single tree by three-dimensional laser scanning, which is characterized in that: firstly, a skeleton model is obtained by processing the point cloud data of a single tree acquired by a three-dimensional laser scanner, and the skeleton model is actually a connected graph; then, Search the nodes connected to each branch according to the topological relationship of the adjacent edges of the skeleton model, extract the branch structure and calculate the branch length; finally, make a cross-section along the skeleton model at the branch point, and use the convex hull algorithm to extract the cross-section The point cloud wraps polygons and computes branch diameters.

Description

A method of three-dimensional laser scanning to determine the branch structure of a single tree

1. Technical field

The invention is a method for obtaining single tree branch structure by processing point cloud data by using a three-dimensional laser scanner.

2. Technical Background

Since the 1980s, with the development of disciplines such as virtual reality, computer graphics, and geographic information systems, as well as the substantial improvement in computer performance, ecologists, agricultural and forestry researchers, and computer researchers have begun to pay attention to the three-dimensional geometry of plants (trees). The model is used to simulate its growth process, describe various characteristic variables, and express the interaction between the environment and trees.

Three-dimensional laser scanning technology has developed rapidly in the last 20 years. Three-dimensional laser scanning or lidar can continuously and rapidly conduct non-contact measurement of the observed object to obtain the distance and reflection intensity from the object surface to the scanner. Laser scanning obtains three-dimensional information on the surface of a large number of objects, also known as point cloud data. Using point cloud data obtained by laser scanners to measure and extract geometric and physiological parameters of plant growth and use them to construct 3D models has become a research hotspot in recent years.

Acquisition of branch structure parameters of trees has always been the research and focus of stock volume estimation and 3D visualization simulation. The main parameters include branch length, branch diameter, branch angle, azimuth, tree height, live branch height and crown width, etc. Width and tree height can be directly measured by point cloud images, and branch length, branch diameter and branch angle can be estimated by extracting single tree point cloud topology as parameters that determine the topological characteristics of trees. representative. The base diameter of branches is an important parameter for visual simulation of tree crowns, which can reflect the accumulation of biomass and the spatial variation characteristics of branches. The point cloud skeleton model can intuitively reflect the morphological structure and topological information of a single tree, and plays an important role in the acquisition and research of morphological parameters.

3. Contents of the invention

A method for measuring the branch structure of a single tree by three-dimensional laser scanning is characterized in that:

First, process the single-wood point cloud data acquired by the 3D laser scanner to obtain a skeleton model. The skeleton model is actually a connected graph. The principle is: voxelize the scattered single-wood point cloud, and divide the point cloud into equal-length cube units , the topological adjacency relationship of cube units is constructed by clustering method, and the units with similar structures are clustered into unit units by weighting according to custom rules. Adjacent units are connected by straight line segments, and each edge connects two vertices with a There is a direction mark, and the position of the vertex is the center of gravity of the scattered points in the voxel;

Then, search the nodes connected to each branch according to the topological relationship of the adjacent edges of the skeleton model, extract the branch structure and calculate the branch length. The principle is: (1) take the lowest point of the trunk as the base point; Search for branch nodes. When a node has more than three adjacent edges, it is determined that the node is a branch node; (3) Each node has a direction attribute. According to the extension direction of the node and the angle between the adjacent edges of the node (two adjacent The size of the skeleton line space is not greater than the angle value of 180°) to determine the branches of different levels;

Finally, make a cross-section along the skeleton model at the branch point, apply the convex hull algorithm to extract the point cloud outsourcing polygon of the cross-section and calculate the branch diameter. The principle is: the skeleton model is a collection of directed line segments, and the calculated skeleton The model has the characteristics of position centrality. The extension direction of the line segment obtained by the least squares fitting is used to extract the point cloud image of the cross-section of a single tree branch, and the point cloud outline is calculated by the convex hull algorithm to extract the branch diameter.

Fourth, the specific implementation method:

1. Process the single-wood point cloud data obtained by the 3D laser scanner to obtain the skeleton model, which is actually a connected graph;

2. Search the nodes connected to each branch according to the topological relationship of the adjacent edges of the skeleton model, extract the branch structure and calculate the branch length;

3. Make a cross-section along the skeleton model at the branch point, apply the convex hull algorithm to extract the point cloud outsourcing polygon of the cross-section and calculate the branch diameter.

Claims (1)

1. a method for three-dimensional laser scanning to measure single wood branch structure, is characterized in that: First, process the single-wood point cloud data acquired by the 3D laser scanner to obtain a skeleton model. The skeleton model is actually a connected graph. The principle is: voxelize the scattered single-wood point cloud, and divide the point cloud into equal-length cube units , the topological adjacency relationship of cube units is constructed by clustering method, and the units with similar structures are clustered into unit units by weighting according to custom rules. Adjacent units are connected by straight line segments, and each edge connects two vertices with a There is a direction mark, and the position of the vertex is the center of gravity of the scattered points in the voxel; Then, search the nodes connected to each branch according to the topological relationship of the adjacent edges of the skeleton model, extract the branch structure and calculate the branch length. The principle is: (1) take the lowest point of the trunk as the base point; Search for branch nodes. When a node has more than three adjacent edges, it is determined that the node is a branch node; (3) Each node has a direction attribute. According to the extension direction of the node and the angle between the adjacent edges of the node (two adjacent The size of the skeleton line space is not greater than the angle value of 180°) to determine the branches of different levels; Finally, make a cross-section along the skeleton model at the branch point, apply the convex hull algorithm to extract the point cloud outsourcing polygon of the cross-section and calculate the branch diameter. The principle is: the skeleton model is a collection of directed line segments, and the calculated skeleton The model has the characteristics of position centrality. The extension direction of the line segment obtained by the least squares fitting is used to extract the point cloud image of the cross-section of a single tree branch, and the point cloud outline is calculated by the convex hull algorithm to extract the branch diameter.