基于曲率特征的点云快速简化算法

为了提高实体反求的效率,提出一种点云快速简化算法.该算法依据特征点群曲率变化的特点在点云邻域拟合曲面上搜寻特征点并进行储存,依据搜寻结果对点云进行特征点分布评估,并根据评估结果设定相应的简化距离对点云进行简化.算法充分保留了特征区域点云,使得简化后的点云能够较好地表达形状,整个搜寻过程只针对高斯曲率极值点的附近点,相对于需要在全局上进行曲率计算的传统简化算法,该算法在运行速度上具有明显优势.

Fast data point simplification algorithm based on curvature character

To improve the efficiency of entity reverse building, the authors proposed an algorithm to simplify cloud data quickly. This algorithm searched and preserved characteristic point according to a rule of curvature change between characteristic point on a surface constructed from a spatial point and its nearest neighbors. Based on the search result, the algorithm gave a characteristic point distributing evaluation to the whole cloud data; and according to the evaluation result, a shortened distance was set to simplify the cloud data. Because the algorithm adequately preserved scattered points in the characteristic area, the cloud data can better express shape after simplification. For the reason that the whole searching process only aims at the maximal point of Gaussian and its neighbors without computing every metrical point, the algorithm greatly raises running speed compared with the traditional cloud data simplification algorithm.