基于自适应邻域大小的点云去噪法  

A point cloud denoising method based on adaptive neighborhood size

作　　者：李沛澄 万程辉[1] 李凤慧 喻文杰 钱铄 LI Peicheng;WAN Chenghui;LI Fenghui;YU Wenjie;QIAN Shuo(School of Hydraulic and Ecological Engineering,Nanchang Institute of Technology,Nanchang 330099,China;Jiangxi Water Investment Engineering Consulting Group Co.Ltd,Nanchang 330099,China)

机构地区：[1]南昌工程学院水利与生态工程学院,江西南昌330099 [2]江西省水投工程咨询集团有限公司,江西南昌330099

出　　处：《南昌工程学院学报》2023年第4期87-91,共5页Journal of Nanchang Institute of Technology

基　　金：江西省研究生创新专项资金项目(YC2022-s980);江西省教育厅科学技术研究项目(GJJ190944)。

摘　　要：针对目前常用点云去噪方法在去噪过程中容易产生过度滤波或滤波不足的现象,提出了一种基于自适应邻域大小的点云去噪法。利用统计滤波去除大尺度噪声,并用主成分分析法求出点云法向量,再通过3个特征值构建局部邻域信息熵函数,依据邻域熵值最小原则判断最优邻域和对应邻域下的法向量,最后通过双边滤波算法进行滤波去噪。实验结果表明,与传统双边滤波算法相比,该方法不仅能达到较好去噪效果,还可保留点云特征,避免过度光顺现象产生。In view of the problem of excessive or insufficient filtering in the denoising process using commonly used point cloud denoising methods,we propose a point cloud denoising method based on adaptive neighborhood size.Firstly,the statistical filtering is used to remove large-scale noise.Then principal element analysis method is used to obtain the point cloud normal vector.A local neighborhood information entropy function is constructed using three eigenvalues,and the optimal neighborhood and corresponding normal vectors are determined based on the principle of minimizing neighborhood entropy.Finally,filtering denoising is performed using a bilateral filtering algorithm.The experimental results show that compared with traditional bilateral filtering algorithms,this method not only achieves good denoising effect,but also preserves the features of point clouds,avoiding the occurrence of excessive smoothing phenomenon.

关 键 词：点云去噪 信息熵 最优邻域估计 法向量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]