基于GA算法的移动曲面拟合在LiDAR滤波中的应用  

Based on Genetic Algorithm Moving Curved Fitting Algorithm for the LiDAR Filtering

作　　者：吴栋 甄宗坤 WU Dong;ZHEN Zongkun(Suzhou Industrial Park Surveying,Mapping and Geoinfomation Co.Ltd.,Suzhou 215000,China)

机构地区：[1]苏州工业园区测绘地理信息有限公司,江苏苏州215000

出　　处：《测绘与空间地理信息》2018年第11期58-61,共4页Geomatics & Spatial Information Technology

摘　　要：作为一种新兴的主动式遥感技术,机载LiDAR可以快速有效地获取目标区域的数字表面模型。在数字城市建设中,往往需要对其中的建筑物进行提取和重建,而建筑物特征提取的关键则是数据滤波,实现对DSM中地面点与非地面点的分类。为建立一种更为有效、稳健的滤波算子,本文拟探讨一种基于遗传算法的移动曲面拟合滤波模型,以克服传统的最小二乘准则对小数据窗口的测量误差明显的放大弊端;另外,通过引入坡度局域标准差,解决在较大地物中心因窗口尺寸较小带来的对地面点的误判,提高模型的适用性。结合相关的实验分析,检验其应用于LiDAR滤波的实用性和可靠性,为后续建筑物的三维重建提供保障。As a new, active remote sensing technology, airborne LiDAR can get the digital surface model fast and effectively from the target area. In digital urban construction, it is often required to extract and reconstruct the buildings from DSM, and the key to the building feature extraction is data filtering, that is to separate the ground point. In order to establish a more effective and robust filter operator, this paper discussed a filtering model of moving curved fitting algorithm that is based on the genetic algorithm, in order to overcome the obvious amplification of the measurement error on a small data window by least square criterion, in addition, through bringing the local standard deviation of slop, we can solve the misjudgment of the ground point owing to the window size smaller in the larger features center and improve the applicability of the model. What’s more, according to related experimental analysis, this paper has tested the practicability and reliability of the LiDAR filtering, which can provide security for the following three-dimensional reconstruction of the buildings.

关 键 词：机载LIDAR 建筑物提取 移动曲面拟合 遗传算法 

分 类 号：P237[天文地球—摄影测量与遥感]