利用LIDAR数据及高分辨率影像的建筑物信息提取

Building extraction from lidar data and high resolution images

文章综合利用LIDAR数据和高分辨率影像互补的特点来提取建筑物信息:在数据预处理,去除LIDAR点云中粗差点,栅格化LIDAR数据以及填补缺失数据的基础上,采用归一化植被指数和渐进形态学方法滤除非地表点,生成完整的数字表面模型;然后分步去除归一化数字表面模型中的非建筑物屋顶点;最后在形态学腐蚀运算结合连通性分析获取种子区域的基础上,生长出完整的建筑物斑块,进而边缘追踪得到建筑物外边缘线。实验结果表明可以有效提取建筑物信息。

This paper proposes a new method for automatic extraction building from LIDAR data andhigh resolution images, the main contents are summarized as follows. （1） Image inpainting is per-formed on the orthorectified aerial imagery, which can reduce the impact of the missing image infor-mation. Then we do the LiDAR preprocessing in preparation for the subsequent filter process. Finallywith the help of the progressive morphological filter and the normalized difference vegetation index,we remove all of the nonground measurements in LIDAR data. （2） We normalize the digital surfacemodel, remove roads, vegetation with other non--building information, and extract the buildingplaques with the seeds which obtained from the morphological erosion operation and connectivity anal-ysis. Experimental results confirm the ability of the presented system to effectively and steadily ex-tract buildings.