Abstract: Aiming at the problems that different scale noise exists in denoising and smoothing of 3D point cloud data and time consuming of algorithm, the denoising algorithm for point cloud data based on noise classification is proposed. According to the distribution characteristics, the noise points are divided into large-scale and small-scale noise. Firstly, the large-scale noise is removed by statistical filtering and radius filtering. Then the small-scale noise is smoothed with fast bilateral filtering. Finally, the purpose of denoising and rapid smoothing for 3D point cloud data are achieved. Compared with the traditional bilateral filtering, the computing efficiency is improved using fast bilateral filtering to smooth the point cloud data. The experimental results show that the proposed algorithm can fast denoise and smooth for 3D point cloud data, which can effectively maintain the geometric features of the scanned object.

Key words: point cloud denoising, fast bilateral filtering, statistical filtering, radius filtering, smoothing

摘要： 针对三维点云模型数据在去噪平滑过程中存在的不同尺度噪声和算法计算耗时问题，提出了点云模型的噪声分类去噪算法。该算法根据噪声点分布特性，将其分为大尺度和小尺度噪声，先利用统计滤波结合半径滤波去除大尺度噪声；然后使用快速双边滤波对小尺度噪声进行平滑，实现点云模型的去噪和平滑。与传统的双边滤波相比，利用快速双边滤波对点云模型数据进行平滑，有效地提高了计算效率。实验结果表明，该算法对点云噪声进行快速平滑去除的同时又能有效地保持被扫描物体的几何特征。

关键词: 点云去噪, 快速双边滤波, 统计滤波, 条件滤波, 平滑