摘要
去噪是点云数据处理中的重要过程,PCL中常用的点距离统计去噪算法涉及邻域规模K和方差倍数m两个参数,不同的取值对去噪结果的影响比较大。在统计去噪原理的基础上,根据正态分布的概率密度函数和3σ准则,对m进行理论分析。其中K值的大小主要影响去噪细节。实验中采用遥感数据进行验证,并通过不同参数组合进行比较。实验结果显示,如果当m大于2时,概率密度比较大,去噪作用比较小;K值对去噪结果的影响不明显。该统计去噪算法可以根据不同的参数值进行不同程度的去噪。
Denoising is an important part in point cloud data processing. The point neighborhood statistics filter in Point Cloud Library is a common algorithm, which involves two parameters the neighborhood size K and the variance multiplier m. Both of the values influence the denois- ing results. Based on statistical denoising principle, carries out an analysis on the m, according to the probability density function of nor- mal distribution and 3σ criterion. Uses the K value decide the details. Remote sensing data in experiments, and makes the comparison under different parameters. The results show that if when m is greater than 2, the probability density is larger, the denoising effect will be small, but K＇s effect is not obvious. The point neighborhood statistics filter can be used for different levels of denoising under different parameter values.
关键词
点云去噪
距离统计
3σ准则
PCL
Point Cloud Denoising
Statistical Distance
3σ Criterion
Point Cloud Library
