基于电塔快速提取的点云配准算法研究

Research on Point Cloud Registration Algorithm Based on Fast Extraction of Electric Tower

摘要: 随着技术发展，基于无人机航摄与激光扫描的电网数据采集方式在电网建设中，尤其是电力巡线、树障分析等工作中具有越来越重要的地位。同时随着数据采集设备的更新换代，点云数据量点数也越来越大。本文针对倾斜密集点云与激光扫描点云在点云配准时点云数量过大的问题，提出一种体素化点云简化的方法。通过将点云体素化后计算出的体素重心来代表体素从而简化点云，利用简化后的点云结合快速点特征直方图(FPFH)算法与迭代最邻近点算法(ICP)实现电塔数据的快速配准。实验表明此方法在不损失配准精度的情况下，有效的提高了配准的速度。

Abstract: With the development of technology, power network data acquisition based on UAV aerial photo-graphy and laser scanning plays an increasingly important role in power network construction, especially in power line patrol and tree barrier analysis. At the same time, with the updating of data acquisition equipment, the number of point cloud data points is also growing. Aiming at the problem that the data volume is too large during the registration of dense matched point clouds or LiDAR point clouds, a voxel-based point cloud simplification method is proposed in this paper. The point cloud is simplified by representing the voxel by its barycenter computed from the point cloud. The fast registration of tower data is realized by combining the simplified point cloud with the fast point feature histogram (FPFH) algorithm and the iterative nearest neighbor point algorithm (ICP). Experimental results show that this method can effectively improve the registration speed without losing the registration accuracy.