基于3D点云的电缆自适应多尺度去噪算法研究

Research on cable adaptive multi-scale denoising algorithm based on 3D point cloud

针对所采集的原始3D点云电缆数据存在大尺度离群点和小尺度起伏噪点的问题,设计出一种基于3D点云的电缆自适应多尺度去噪算法。首先,对于原始3D点云电缆数据应用统计离群消除滤波算法滤波,去除大尺度空间孤立离群点;其次,对于传统双边滤波的两个参数σc以及σs应用了基于半径和标准差的自适应改进;最后,采用提出的基于半径和标准差的自适应双边滤波算法去除3D点云电缆数据中的小尺度起伏噪点。实验结果表明,提出算法对3D点云电缆数据能多尺度去除噪点,且无需人工调整参数,明确了电缆边界,实现了理想的点云去噪效果,为后续电缆分割、识别等可视化巡检奠定了良好基础。

Aiming at the problems of large scale outliers and small scale fluctuation noise points in the original 3 Dpoint cloud cable data collected,an adaptive multi-scale de-noising algorithm for cables based on 3 Dpoint cloud was designed.First of all,the original 3 Dpoint cloud cable data is filtered by the statistical outlier elimination filtering algorithm to remove isolated outliers in large-scale space.Secondly,adaptive improvements based on radius and standard deviation are applied to the two parameters C and S of traditional two-sided filtering.Finally,the proposed adaptive bilateral filtering algorithm based on radius and standard deviation is used to remove the small scale fluctuation noise points in 3 Dpoint cloud cable data.Experimental results show that the algorithm proposed in this paper can remove noise points at multiple scales on 3 Dpoint cloud cable data without manual adjustment of parameters,define the cable boundary,achieve ideal point cloud denoising effect,and lay agood foundation for subsequent visual inspection of cable segmentation and identification.