基于第二代Curvelet变换的散乱点云去噪方法

提出了一种新的针对无组织散乱点云的去噪处理方案:首先,建立三维栅格阵列,并对散乱点进行空间栅格划分,初步确定散乱点之间的三维拓扑位置;第二步,运用最小二乘法确定栅格中散乱点的最佳逼近平面,以平面的法向量为z轴建立新的三维坐标系,并将栅格内的散乱点向该坐标系进行坐标转换。此方法对散乱点的深度信息具有很好的分离效果;第三步,将第二代Curvelet变换的最新研究进展引入到文章的方案中,对投影后的散乱点当做图像型数据进行Curvelet变换,设置阀值对变换后的系数进行筛选,从而达到去除噪声的目的。文中所采用的方案充分利用了无组织散乱点云中包含的深度信息,对噪声敏感。从试验的结果来看,获得了满意的散乱点云去噪结果。这些数据也可以作为快速成型制造和逆向工程的数据源。

Unorganized Point Cloud Denoising Method Based on the Second Generation Curvelet Transform

Based on the second generation Curvelet transform,this paper proposes a new denoising method for unorganized point cloud.Firstly,3D grid array is constructed and the unorganized points are partitioned to the different grid according to its' coordinates.Secondly,taking advantage of the method of least square,the best plane approximating to the points in the grid can be obtained.With the normal of approximating plane as the z-axis,a new 3D coordinate system can be built and then the points in the grid are transformed into the new reference.By this measure,the depth informations of unorganized points can be efficiently separated.Finally,by using the second generation Curvelet transform,the Curvelet coefficients are acquired and filtered by appropriate thresholds,and the noises can be eliminated by this way.The experiment shows that the depth informations of unorganized point cloud are efficiently utilized and the denoising efficiency are satisfying.