地面点云数据快速重建建筑模型的分层方法

摘要

为了快速从地面激光扫描数据重建建筑物模型，避免数据处理中复杂的点云栅格化和面片分割，提出一种基于分层方法的建筑三维模型快速重建方法。利用直方图统计方法去除噪声，结合大多数建筑物自身特点，对点云数据进行分层并垂直投影至各层平均高度的平面，得到建筑物各层的边缘特征点，进而规则化，实现建筑物三维数字模型的快速重建。实验结果表明，方法可以快速提取建筑物边界点，简化传统点云数据在建筑物重建中的复杂过程和大量的数据处理运算，而且最终模型的精度可以得到保证，具有一定的普适性。%T he objectives of this research aim to present a layered method to reconstruct building models from Terrestrial Laser Scanning (TLS ) data and to improve the efficiency of building reconstruction greatly .Firstly ,statistic histogram method is used to filter noisy data .Secondly ,due to most buildings with regular features , the points cloud data are layered to different overlapping layers and projected vertically to the average height of each layer , thus the building boundary points of each layer being extracted .Thirdly ,by regularizing these boundary points and connecting all layers ,the building model is established quickly .Experiments show that the boundary points extraction based on layered method can achieve good results for most buildings reconstruction , and it facilitates the complicated procedures of traditional method ,such as points cloud rasterization and planar surface segmentation which will cost lots of operations during reconstruction procedures .