改进的分层曲率海量点云精简算法

IMPROVED MASSIVE POINT CLOUD REDUCING ALGORITHM BASED ON STRATIFIED CURVATURE

海量点云精简既要考虑算法的复杂度,又要考虑精简结果的效果。根据三维扫描仪形成的点云特点,提出将空间点云划分为扫描层平面点云,从而将空间问题转化为平面问题。通过平面内Angl的简单计算获得点曲率,从而简化算法复杂度;通过引进距离参数Dis防止精简"大孔洞"的出现;通过综合考虑点的曲率和点间的距离,形成一个判别点是否被删除的标准,修改该判别标准公式中的系数,可以得到不同的精简效果。试验结果证明,该算法对海量点云的精简实践可行,具有复杂度低、数据精简率高等特点。

Massive point cloud reduction shall consider both the complexity of the algorithm and the effect of reducing result. According to the characteristics of point cloud formed by 3-D scanner,we proposed to divide the spatial point cloud to the planar point cloud on scanning level,so that converted the space problem to the plane problem. Through simple calculation of Angl within the plane we got the point curvature,thus simplified the complexity of the algorithm. By introducing distance parameter Dis we avoided the emergence of＂big hole＂in reduction. By comprehensively considering the curvature of points and the distance between points,we formed a criterion determining whether to delete the points or not,and by modifying the coefficients in formula of determination criterion we could obtain different reduction effects.Test result proved that this algorithm was feasible in reducing practice of massive cloud point,and had the features of low complexity and high data reducing rate.