保留边界特征的点云简化算法

为有效简化点云数据，提出保留边界特征的点云简化算法。该算法利用三维栅格划分法建立散乱点云的空间拓扑关系，计算每个数据点的近邻，通过球拟合法求得其曲率和具有方向性的法向量，采用投影点个数比值法找到并保留点云边界，根据具体情况设定所需阈值，对非边界点进行分类，通过对点的曲率与平均曲率比较、近邻保留点与近邻点个数比例，完成点云简化。实验结果表明：该算法不仅能对点云进行直接有效地简化，而且还能很好地保留点云模型的细节特征，简化比例达25％～40％。该方法可以满足不同种类点云简化的要求，能够提高计算机运行效率。

Research on point cloud simplification with boundary features reservation

This paper proposes a simplification method for point cloud with boundary feature reservation for effective simplification of the point cloud. This algorithm consists of firstly using the 3 D grid sub- division method to represent the spatial topology relationship of the scattered point cloud and calculate the k-nearest neighbors for each data point, using the ball-fitting method to simply compute the curvature and the directional normal vector, and then identifying and reserving all the boundary points according to the ratio of the number of projected points, setting the desired thresholds by the specific situations, and classifying the non-boundary points through these thresholds, and finally simplifying the scattered point cloud according to comparative study of curvature and mean curvature of the points and the proportion of reserved points in their k-nearest neighbors. The algorithm is verified by reducing some typical point cloud cases with various surface features. The experimental results indicate that the algorithm, marked by setting the threshold size according to simplification requirements, allows the direct and effective reduction of point cloud, while preserving detail feature of point cloud model, with a simplification proportion up to 25%-40%. This method can fulfill the requirements for simplifying different point cloud and improve the efficiency of computer operation.