结合超体素与颜色信息的区域生长点云分割方法

Region growing point cloud segmentation method combining supervoxel and color information

为解决传统区域生长点云分割算法存在的欠分割和过分割现象,提出一种结合超体素与颜色信息的区域生长点云分割方法。在分割过程中加入超体素过分割步骤,避免直接从点云中分割数据,有效消除噪声和异常值对分割的影响,利用一种几何和颜色信息的联合准则合并超体素并进行区域生长。与深度学习方法和其它3种传统分割算法相比,分割效率和精度都得到了较大提升,解决了欠分割和过分割的问题。

To solve the problems of under segmentation and over segmentation in traditional region growing point cloud segmentation algorithm,a point cloud segmentation algorithm combining supervoxel and color region growth was proposed.The super voxel over segmentation step was added in the segmentation process,which avoided the direct segmentation of data from the point cloud and decreased the effect of yawp and exception value on segmentation algorithm.A joint criterion of geometric and color information was used to merge the super voxels and grow the region.Compared with the depth learning method and the other three traditional segmentation algorithms,the segmentation efficiency and accuracy are greatly improved,and the problems of under segmentation and over segmentation are well solved.