一种针对大规模场景的点云匹配算法

A Point Cloud Matching Algorithm for Large-scale Scenarios

针对大规模点云匹配时传统算法速度慢和匹配结果不一致的问题,提出一种新的点云匹配方法。该方法首先利用KD树找到点云中深度最小的点并以该点作为种子点,然后通过在深度信息和曲率两个方面做以改进的区域生长分割算法提取出点云上表面区域,并在该区域提取点云边界。最后使用改进的点对特征完成点云匹配算法验证。实验结果表明,相比传统算法,该方法在匹配速度以及匹配结果的一致性方面得到了显著的提升,在处理大规模点云匹配上具有实际应用价值。

A new point cloud matching method is proposed to address the issues of slow speed and inconsistent matching results in traditional algorithms in large-scale point cloud matching.This method first uses the KD tree to find the point with the minimum depth in point cloud and uses it as a seed point.Then,an improved region growth segmentation algorithm improving in depth information and curvature is used to extract the upper surface area of point cloud,and the point cloud boundary is extracted in this area.Finally,the point cloud matching algorithm is validated using improved point pair features.The experimental results show that compared with traditional algorithms,the proposed method has significantly improved matching speed and consistency of matching results,and has practical application value in handling large-scale point cloud matching.