保持细节特征的点云去噪算法

A Denoising Method for Point Cloud with Detail Feature-Preserving

三维点云数据的采集过程受到仪器精度、人为扰动和复杂环境等因素影响,存在大量的噪点并且分布不均匀,造成基于点云数据的重建模型尖锐特征模糊。对此,提出一种保持细节特征的点云去噪算法以提高点云模型的精度。该算法首先通过八叉树建立点云的拓扑关系,加快点云的邻域搜索速度;其次,采用主成分分析法估算点云的法向量;最后,将双边滤波器与加权局部最优投影算法结合,实现对点云的去噪均匀化处理。实验结果表明,所提算法不仅能很好地去除点云数据中的噪点,而且同时保留了点云重建模型的细节特征。

The acquisition process of 3D point cloud data is affected by instrument accuracy, human disturbance, complex environment and other factors. There are a large number of noise points and their distribution is not uniform, resulting in the fuzzy sharp features of the reconstruction model based on point cloud data. Therefore, this paper proposes a point cloud denoising algorithm that preserves the detailed features to improve the accuracy of the point cloud model. Firstly, the algorithm established the topological relationship of point cloud by octree to speed up the neighborhood search of point cloud. Secondly, the principal component analysis method is used to estimate the normal vector of the point cloud. Finally, the two-sided filter and weighted local optimal projection algorithm are combined to realize the denoising and homogenization of the point cloud. The deviation analysis diagram proves that the algorithm can not only remove the noise in the point cloud data well, but also retain the detailed features of the point cloud reconstruction model.