装配面点云特征增强的去噪算法研究

为了去除与法兰装配面数据点混合在一起的噪声并更好地保留装配面的特征,将装配面点云数据中的噪声分为 2 类。利用空间栅格划分将点云数据栅格化,使用 K 领域搜索的方法筛选出第 1 类噪声点,运用采样点邻域点数统计法删除第 1 类噪声点;采用改进的双边滤波法对第 2 类噪声点进行去噪。通过与各类算法进行比较,实验结果表明,该去噪算法在达到预期去噪效果的同时还可以增加装配面特征的保持度,并且能够避免传统双边滤波在去噪后产生光顺过度的现象。

Research on Denoising Algorithm for Reature Enhancement of Assembly Surface Point Cloud

To retain the features of flange assembly surface when the noise in the data of the flange assembly surface is removed,the noise in the assembly surface point cloud data are divided into two categories.The point cloud data is formatted by using space division;the first noisepointis selected by using the K field research method;the first point is removed through sampling points neighborhood noiselike statistical method;the second type of noise points is denoised with the improved bilateral filtering method.Compared with various algorithms,the experimental results show that the denoising algorithm described herein can improve the retention of assembly surface characteristics while meeting the expected effect,and can avoid the phenomenon of excessive smoothing caused by denoising traditional bilateral filtering.