一种新的三维点云模型去噪光滑算法研究  

RESEARCH ON A NEW DENOISING ALGORITHM FOR 3D POINT CLOUD MODEL

作　　者：廖昌粟 何东建[2] 王美丽[1] 

机构地区：[1]西北农林科技大学信息工程学院,陕西杨凌712100 [2]西北农林科技大学机械与电子工程学院,陕西杨凌712100

出　　处：《计算机应用与软件》2017年第4期281-287,共7页Computer Applications and Software

基　　金：国家高技术研究发展计划项目(2013AA102304)子课题;第56批中国博士后基金项目

摘　　要：提出一种三维点云模型的去噪光滑算法。该算法根据密度滤波和点法矢量信息对离群噪声点进行去除;再利用张量投票算法和数据点的近邻点在其最小二乘平面上投影的分布均匀性检测出模型的边界特征,并对特征实现加强操作;最后,采用双边滤波对模型表面进行光滑。实验表明,该算法能有效地对模型进行去噪光滑处理,且由于对模型边界特征进行了保留与加强,从而避免了模型光滑操作对模型特征造成损害的问题。This paper proposes a denoising algorithm for 3D cloud model. The algorithm removes outlier noise point based on density filtering and vector information of points. Then, the boundary feature of the model is detected by the tensor voting algorithm and the uniformity of the projection of the nearest neighbors of the data points on the least squares plane, and the features can be enhanced. Finally, the model surface is smoothed by bilateral filtering. The experimental results show that the algorithm can effectively smooth the model and preserve and strengthen the boundary feature of the model to avoid the problem that the smooth operation of the model damages the model features.

关 键 词：三维点云 边界检测 密度滤波 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]