基于噪声分类的双边滤波点云去噪算法

发布时间：2017-12-31 15:03

本文关键词：基于噪声分类的双边滤波点云去噪算法　出处：《计算机应用》2015年08期 　论文类型：期刊论文

【摘要】：针对三维点云数据模型在去噪光顺中存在不同尺度噪声的问题,提出一种基于噪声分类的双边滤波点云去噪算法。该算法首先将噪声细分为大尺度和小尺度噪声,并使用统计滤波结合半径滤波对大尺度噪声进行去除;然后对三维点云数据进行曲率估计,并对现有点云双边滤波进行改进,增强其鲁棒性和保特征性;最后使用改进的双边滤波对小尺度噪声进行光顺,实现三维点云数据模型的去噪、光顺。与单独使用双边滤波、Fleishman双边滤波相比,改进算法在三维点云数据模型光顺平均误差指标上分别降低了50.53%和21.67%。实验结果表明,该改进算法对噪声进行尺度的细分既提高了计算效率,又避免了过光顺和细节失真,较好地保持模型中的几何特征。
[Abstract]:Aiming at the problem of 3D point cloud data model in the light of different scale noise denoising fairing, proposed a bilateral filtering denoising algorithm based on cloud point noise classification. Firstly, the noise is divided into large and small scale noise, and use statistical filtering to remove noise with the large scale and the radius of filtering; the three-dimensional point cloud data of curvature estimation, and the existing point cloud of bilateral filtering is improved, enhanced the robustness and security of characteristic; finally, the use of improved bilateral filtering on small scale noise smoothing, 3D point cloud data denoising, smoothing and using bilateral filtering. Compared with Fleishman alone, bilateral filtering the improved algorithm, the fairing in the 3D point cloud data model of average error index were decreased by 50.53% and 21.67%.. The experimental results show that the improved algorithm for segmentation of noise scale calculation is improved. Rate, and avoid the excessive smoothing and detail distortion, preserve the geometric features of the model.

【作者单位】：桂林电子科技大学信息与通信学院;
【基金】：广西自然科学基金资助项目(2013GXNSFDA019030,2013GXNSFAA019331,2012GXNSFBA053014,2012GXNSFAA053231,2014GXNSFDA118035) 广西教育厅项目(201202ZD040) 桂林电子科技大学研究生教育创新计划项目(GDYCSZ201453)
【分类号】：TN911.4
【正文快照】： 0引言随着三维激光扫描技术的高速发展,人们可以很容易获取真实世界中物体的表面三维点云数据,因此三维点云数据模型被广泛应用于虚拟现实、逆向工程、城市建模等[1]。然而,受到扫描设备本身精度、人为操作和被测物体表面材质等因素的影响,三维点云数据在获取阶段不可避免存在

【参考文献】