摘要
分析机载LiDAR点云与影像数据特点,提出了一种建筑物点云与配准后影像相结合的建筑物轮廓信息提取方法。首先,采用α-shapes算法从点云中提取粗糙的建筑物轮廓多边形;然后,采用基于线支撑区域的直线段提取算法从影像中提取边缘信息,并利用投票机制,以点到直线的距离为因子,从中过滤出真实的建筑物边界;最后,提出一种建筑物轮廓精化的新方法,利用从影像中提取的边缘信息修正从点云中提取的粗糙轮廓,并对修正后的轮廓采用道格拉斯-普克算法去除冗余节点,采用强制相交方法恢复建筑物转角,最终得到了准确的建筑物外轮廓多边形,并通过实验验证了该方法的有效性。
A method for extracting building boundaries using airborne LiDAR point cloud data and imageries is proposed in this paper. Firstly, an α-shape algorithm is used to extract the rough outline of buildings from point clouds. Then building edge line segments are extracted from the registered images by an straight line segments extraction algorithm based on line region support. By using voting mechanism and point-to-line distance, the true boundaries of the buildings are obtained. Finally, a new method for refinement of a building outline is put forward, in which the extracted edge information is utilized to correct the rough outline extracted by the point cloud image, and the revised outline is processed by Douglas - Peucker algorithm to remove redundant nodes. the force intersect method is employed to restore the corner of the building, and finally the accurate outside contour polygons of the building is obtained. The effectiveness of the proposed method has been verified by experiments.
出处
《国土资源遥感》
CSCD
北大核心
2014年第2期54-59,共6页
Remote Sensing for Land & Resources
基金
国家科技支撑计划项目(编号:2013BAJ04B03)资助
关键词
LIDAR
点云
航空影像
建筑物轮廓提取
LiDAR
point clouds
airborne imageries
building boundary extraction
