关键词: 线阵激光雷达, 推扫成像, 激光点云, 无人机, 目标识别

Abstract: A dynamic imaging model for UAV-based line-array LiDAR is proposed to address the scarcity of point cloud datasets in UAV imaging scene. A pillar-voxel based point cloud target detection algorithm is proposed to verify the authenticity of the simulation data. Establish typical target models and scenarios with rugged terrain, camouflage, and vegetation cover within a virtual simulation platform. Combine point cloud simulation model, UAV motion model, stitching imaging and distortion correction methods to generate the point cloud dataset. Annotate the dataset using a completeness judgment method based on overlapping areas. Use a pilar-voxel feature extraction module to process the target's top features. Train the point cloud target detection algorithm using the annotated simulation dataset, and evaluate the algorithm on a real dataset obtained from equivalent experiments. The detection algorithm achieves an accuracy rate of 93.2% on the real dataset. The evaluation result indicates that the simulation data effectively reflect the true characteristics of the targets, and the dynamic imaging model has high credibility.

Key words: line-array LiDAR, pushbroom imaging, laser point cloud, unmanned aerial vehicle, target recognition