影像辅助下LiDAR数据建筑物轮廓信息提取

Building boundary extraction using LiDAR data and images

针对复杂场景中提取的建筑物外轮廓锯齿状变形及精度不高等问题,提出了一种新的轮廓信息提取方法。在综合利用LiDAR数据及影像特征的基础上,建立轮廓线提取候选区,计算候选点的方向、梯度及高边点信息,引入基于活动轮廓的图割算法(graph cuts based active contour model,GCBAC)构建并解算轮廓线能量函数,实现建筑物轮廓信息的精确提取。通过3个具体实例证明了该方法的有效性和可行性。

In this paper, the strategy to extract accurate building boundary from LiDAR data and images was explored. The workflow is as follows： first LiDAR data and images feature are used to extract building blobs. Then contour extraction candidate regions are established, and gradient and direction information of the candidate points are calculated to build the classic energy function. Finally energy function is computed with GCBAC algorithm, and the building boundary will be generated after the iterative optimized approach. The three experiments show that the strategy proposed in this paper is an effective method.