Unlock instant, AI-driven research and patent intelligence for your innovation.

Single-plant tree crown segmentation method based on deep learning and airborne laser point cloud

An airborne laser and deep learning technology, applied in the field of forestry, can solve the problem of losing three-dimensional space position information

Pending Publication Date: 2021-05-18
NANJING FORESTRY UNIV
View PDF0 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The method based on multi-view has achieved good results in classification tasks, but the original 3D spatial position information will be lost in the process of converting to 2D images

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Single-plant tree crown segmentation method based on deep learning and airborne laser point cloud
  • Single-plant tree crown segmentation method based on deep learning and airborne laser point cloud
  • Single-plant tree crown segmentation method based on deep learning and airborne laser point cloud

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0081] The specific embodiment of the present invention is described further below according to accompanying drawing:

[0082] The present embodiment provides a single tree canopy segmentation method based on deep learning and airborne laser point cloud, which mainly includes: (1) collecting data using unmanned aerial vehicle-borne laser LiDAR; (2) voxelizing the training and testing sites ; (3) Convert the data of the training and testing sites from voxelization to the format required by the model PointNet for training and testing; (4) Identify the segmented voxels based on the model PointNet, and use the gradient information to construct and describe the structure of each voxel tree Boundary, to achieve the division of a single tree canopy. The workflow of the method of the present embodiment is as follows figure 1 shown.

[0083] Research plot:

[0084] The research area is located in the Qishan Scenic Area of ​​Chizhou City in the southwest of Anhui Province (30°38’15.8...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a single-plant tree crown segmentation method based on deep learning and airborne laser point cloud. The method comprises the following steps: acquiring point cloud data of a research site; dividing the denoised point cloud data into overground points and ground points; extracting point clouds of single trees in the ground points, and dividing different trees into different voxels; constructing a training sample data set; carrying out training on the PointNet deep neural network; subdividing an overground point of a to-be-tested site into a plurality of voxels through a voxelization method, converting point cloud data in the voxels into a format required by PointNet, inputting the point cloud data into a trained PointNet model, and identifying a point cloud in each voxel of the tree; and positioning boundary points of the crown by combining the gradient information of the DSM of each voxel with an inertial momentum gradient method, and drawing the segmented crown according to the boundary points. According to the method, the trees are identified on the voxel scale, the single-plant crown is defined in combination with the highly-related gradient information, and the single-plant crown segmentation accuracy is high.

Description

technical field [0001] The invention belongs to the technical field of trees, and in particular relates to a single tree crown segmentation method based on deep learning and airborne laser point cloud. Background technique [0002] Accurate separation of individual trees plays a crucial role in tree parameter inversion. Forest parameters, such as tree location, tree height, canopy density, tree crown width, tree species, and diameter at breast height, are critical for forest resource management, field inventory, and silvicultural activity execution. Traditional tree structure parameters are usually obtained through on-site measurement, but this process is very time-consuming, labor-intensive, and destructive. Light detection and ranging (LiDAR), an automatic remote sensing technology, has become one of the most effective measurement techniques for obtaining detailed and accurate target phenotype data due to its high precision and high efficiency. According to different del...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/10G06T5/00G06K9/62G06N3/04G06N3/08
CPCG06T7/10G06N3/08G06T2207/20081G06T2207/20084G06T2207/30188G06T2207/10028G06T2207/10044G06N3/047G06F18/241G06T5/70
Inventor 云挺陈鑫鑫张运玲曹林
Owner NANJING FORESTRY UNIV