Street tree target recognition method based on vehicle 2D LiDAR point cloud data

Abstract

The present invention discloses a street tree target recognition method based on vehicle 2D LiDAR point cloud data. The method comprises: acquiring street data by using a vehicle 2D LiDAR system, saving the street data as a point cloud file, and performing crown and non-crown category labeling on the point cloud file; constructing a variable scale grid, and extracting the three-dimensional spherefrom the variable scale grid; calculating point cloud eigenvectors in the three-dimensional sphere to obtain a point cloud eigenvector set; learning a crown point cloud classifier from the point cloudeigenvector set through an SVM algorithm; and according to the crown point cloud classifier, performing online recognition on point cloud frames to obtain a spray prescription map. According to the method disclosed by the present invention, the vehicle 2D LiDAR which with small data redundancy and easy-for-online processing performance is used to obtain urban street data, and to provide an accurate spray basis for applying pesticide to the target; and a point cloud sequence spatial index structure that preserves data accuracy, improves neighborhood retrieval efficiency, and meets online processing requirements is established, and accurate recognition of street tree targets is realized.

Description

technical field [0001] The present invention relates to the field of target recognition, and more specifically, to a street tree target recognition method based on vehicle-mounted 2D LiDAR point cloud data. Background technique [0002] Street trees are an important part of urban ecosystems and urban landscapes. Facing the increasingly serious environmental pollution problems, street trees are of great significance in improving urban ecological environment, purifying air, regulating climate and conserving water sources. Affected by the greenhouse effect, human disturbance and other factors, the pests and diseases of street trees are increasing day by day, resulting in the withering or death of street trees, which not only seriously affects the greening and beautification effect of street trees, but also directly affects the urban ecological environment and residents' lives, becoming a restrictive factor for fine management of landscaping . [0003] Due to the large spacing ...

AI Technical Summary

Problems solved by technology

Uniform spatial index structures such as grids and voxels can effectively reduce the amount of point cloud data and data redundancy, and improve the efficiency of neighborhood retrieval, but it is difficult to deal with uneven point cloud densities. At the same time, grid / voxel size selection and segmentation Accuracy is difficult to balance

Spatial index structures such as quadtrees, octrees, and KD trees can better adapt to the uneven distribution of data, retain data accuracy, and facilitate batch display and storage of point clouds, but they are not suitable for online data processing.

Images