基于多视点云数据融合算法研究  

A Fusion Algorithm Based on Multi View Point Cloud Data

作　　者：惠广裕[1] 何红丽[1] 

机构地区：[1]中国飞行试验研究院,陕西西安710089

出　　处：《北京测绘》2015年第3期48-50,47,共4页Beijing Surveying and Mapping

基　　金：航空科学基金项目(2012ZD30)

摘　　要：针对光学测量中多幅点云的数据融合,提出了一种基于法向滤波的多幅点云融合算法.该算法首先对多幅点云法向滤波,通过2幅点云的双向查找来寻找种子点,在点的法向方向寻找2幅点云中对应的k邻域,计算邻域点的加权和,而融合点是种子点沿其法向移动的结果。与平均聚类法相比,该方法获得的模型表面更加光顺,特征更明显,点的分布也更均匀,对于包含粗大匹配误差的多幅点云模型的融合具有较好的效果.For the fusion of multiple pieces of point cloud data in optical measurement, puts forward a method to filter fusion algorithm based on multi range images. The algorithm first filters multiple point cloud normal, to search for the seed points by 2 point clouds of the bidirectional search, at the point of the law to the direction finding 2 point cloud the K neighborhood should, weighted and the points in the neighborhood, and the fusion point is the seed point along its normal mobile results. Compared with the average clustering method, the surface model obtained by this method is more smooth, more obvious characteristics, point distribution is more uniform, and has better effect on fusion a plurality of point cloud model, containning coarse matching error.

关 键 词：法向约束 多幅点云 数据融合 

分 类 号：P208[天文地球—地图制图学与地理信息工程]