上海科技大学知识管理系统

本发明的技术方案是提供了一种基于球谐变换和迭代优化的全局配准方法，其特征在于，假设：第一、预定义输入点云中正确匹配对的最小百分比，表示为点云离群值的数量限制；第二、用于确定正确匹配对的距离阈值是由场景预设的，表示为点云离群值的限制距离。在本发明提供的算法中，点云先经过粗配准得到若干搜索域，然后使用评估标准对于搜索域进行排序，使用分支定界法排除不正确的搜索域并得出最终的配准结果。本发明公开的算法可以应用于三维重建。由于在相同精度下本发明的算法比目前所有的全局配准算法更快，所以对三维重建在相同时间内，本发明能达到比目前所有算法更高的精度。

Provided in the technical solution of the present invention is a global registration method based on spherical harmonic transform and iterative optimization. The method is characterized by assuming that : first, a minimum percentage of correct matching pairs in input point clouds is predefined, and is expressed as a quantitative limitation of point cloud outliers; and second, a distance threshold value used for determining the correct matching pairs is preset by a scenario, and is expressed as a limited distance of the point cloud outliers. An algorithm provided by the present invention allows point clouds to undergo coarse registration to obtain a plurality of search domains, then sorts the search domains by using evaluation standards, and excludes incorrect search domains by using a branch and bound method so as to obtain a final registration result. The algorithm disclosed by the present invention can be applied to three-dimensional reconstruction. At the same precision, the algorithm provided by the present invention is faster than all current global registration algorithms; therefore, with respect to three-dimensional reconstruction within the same time, the present invention can achieve higher precision than all of the current algorithms.