基于骨架的三维点云模型分割

针对目前三维模型分割方法在分割细节方面准确度低、分割结果不明确的问题,通过研究三维模型骨架数据的拓扑关系,提出一种以提取三维点云模型骨架为预处理步骤的分割算法。提取对三维模型进行骨架提取,通过骨架数据之间的关系,提取骨架关键点;选择同一邻域内关键点到质心的距离最小的点作为最终的关键点,对已找到的关键点进行优化处理,减小错误关键点对分割结果的影响;使用找到的关键点,通过改进的区域生长算法,得到分割后的骨架区域,即可得到相应原始数据的分割结果。实验结果表明,采用该方法得到的分割结果相较于现有分割方法可以准确获得三维模型的语义分割结果及较好的分割边界。

Skeleton-based three-dimensional point cloud model segmentation

To solve the problem of low accuracy and unclear segmentation results of the current 3 dmodel segmentation method,by studying the topological relation of three-dimensional model skeleton data,a segmentation algorithm was proposed which took extracting the three-dimensional point cloud model skeleton as the preprocessing step.The skeleton of the 3 dmodel was extracted,and the key points of the skeleton were extracted through the relationship between the skeleton data.The point with the smallest distance from the key point to the center of mass in the same neighborhood was selected as the final key point,and the key point was optimized to reduce the influence of the error key point on the segmentation results.The segmented skeleton region was obtained by using the key points and the improved regional growth algorithm,and the segmentation result of the corresponding original data was obtained.Results of the experiment show that the semantic separation of the three-dimensional model obtained by using the method is more accurate with better dividing line than that using the existing method of segmentation.