关键词: 堆石坝;检测坑;点云提取;法向量;区域聚类分割;

Abstract: Using 3D laser scanning technology to obtain the point cloud of the rockfill dam inspection pit can quickly calculate the volume of the inspection pit and the compaction quality of the rockfill dam.The key is accurately segmenting the target inspection pit from scanned point cloud data.In this paper, a point cloud extraction method of inspection pit surface based on normal vector region clustering segmentation was proposed.Firstly, the external point P of the rolling surface was detected by the local density method, and the remaining points were the internal point Q;then the least square method was used to fit the outer point plane, and the angle between the normal vector of the fitting plane and the normal vector of the inner point was θ.Finally, the outermost points were used as the seed points, and the search area was gradually expanded until there was a θ greater than the threshold angle within the area.The intersection between the compaction surface and the inspection pit was determined through the above method, so the point cloud of the inspection pit was accurately extracted.The method of rapid extraction of the inspection pit point cloud data was implemented through MATLAB programming.The feasibility of the method was verified by conducting experiments on inspection pits on three different dam materials.The research results can provide reference for point cloud extraction of similar shape targets and rapid detection of compaction quality of rockfill dams.

Key words: rock dam; inspection pit; point cloud extraction; normal vector; region clustering segmentation;