特征保留的点云数据自适应精简算法

摘要

提出一种特征保留的点云数据自适应精简算法.该算法首先构造散乱点云数据的局部拓扑信息,通过一种改进的二次栅格法快速建立K邻域,由此估算点的邻域弯曲度,再进行分类.算法在保留特征点后对其余点应用自适应精简距离进行阈值精简,故算法不仅可以完整保存实物模型整体轮廓,而且能够最大限度地保证模型区域特征.数值实验结果表明,该算法能够得到不错的精简效果,且具有较小的计算时间复杂度.%In this paper, an adaptive simplification method for point cloud data with feature reservation is presented. In this algorithm, we first construct local topology information of scattered point cloud data, and fast establish K-Nearest Neighbours with an improved quadratic grid method, and estimate via this the neighbourhood curvature of the points, and then make the classification. In the algorithm, after the feature points are reserved, we simplify the threshold value against remaining points using adaptive simplification distance, therefore the algorithm can entirely preserve the whole contour of real entity model and to guarantee on the maximum the regional features of the model. Numerical experiment result in the paper proves that this algorithm is able to achieve quite good simplification effect with less complexity in computation tune.