基于法向修正及中值滤波的点云平滑  

Denoising of Point-Sampled Model Based on Normal Mollification and Median Filtering

作　　者：苏志勋[1] 栗志扬[1] 王小超[1] 

机构地区：[1]大连理工大学数学科学学院,大连116024

出　　处：《计算机辅助设计与图形学学报》2010年第11期1892-1898,共7页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(60673006;U0935004)

摘　　要：为了在降噪的同时保持点云的几何特征,并较好地处理离群点,通过对传统的法向量修正方法进行改进,提出基于L1中值滤波的点云平滑方法.根据核回归估计理论建立一种法向修正的框架,并在此框架下引入二阶Hessian矩阵,提高了法向估计的精度;以经典的L1中值理论为基础,提出一种迭代降噪的方法,并指出了该方法与mean-shift方法和双边滤波方法的联系.实验结果表明,该法向修正及点云平滑方法简单、有效,在处理原始点云时具有优势.In this paper we propose a denoising algorithm based on L1 median filtering for noisy sets of points.The proposed algorithm improves the traditional methods to better preserve the surface features and handle well the outliers.Based on the kernel regression theory,we establish a framework for normal mollification.The framework utilizes the Hessian matrix and predicates more accurate normals.Using the framework we present a novel iterative denoising method for point-sampled model based on the classic L1 median theory.The comparison of our method to the classical mean-shift denoising method and bilateral filtering method are discussed.The experimental results are presented which show the efficiency of the algorithm and the simplicity of implementation.

关 键 词：核回归 L1中值 几何光滑 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]