基于AVP的不均匀散乱点云离群点去噪算法

AVP-based outliers removed algorithm in uneven scattered point cloud

为去除采空区点云数据噪声,解决去噪时点云模型孔洞扩大和几何特征弱化等问题,提出一种基于逼近视平面(approximate view plane,AVP)的离群点去噪算法。规则网格化点云数据,基于邻域网格重心构造中心网格的AVP;根据网格内点到AVP距离,自适应确定阈值剔除噪声,二次迭代调整网格尺寸提高噪声检测率。实验结果验证了该算法对不均匀散乱点云在去噪效果和模型几何特征保持等方面优于传统的离群点剔除算法。

To remove the mined-out area of point cloud data noise,solve the problems of expanded holes of point cloud model and weakened geometric feature,an improved outliers removed algorithm based on approximate view plane（AVP）was presented.The point cloud data were put into specification grid,an approximate view plane based on neighborhood center was constructed according to the grid points.The noise point was removed according to the distance between point and the plane,and iterative adjustments were applied to grid size to eliminate more noise points.Experimental results show that the algorithm for uneven scattered point clouds is better than traditional outliers removed algorithm in terms of denoising effect and keeping the other aspects of geometric features.