一种引入 RANSAC 算法的匹配点云规则化方法

针对影像匹配点云的大数据量和分布不规则化问题，本文提出了一种基于平均模型的RANSAC规则化方法。该方法首先对匹配点云进行格网化，然后以格网为基本单位，利用RANSAC方法重复随机抽取多个样本，根据高程平均模型和设定的判断条件得到最终一致集，最后基于该子集利用平均模型和最邻近原则完成该格网的规则化。对比实验证明，本文方法在保证数据精度的同时，大大提高了数据的简化度和可靠性。

A Method Adopting RANSAC Algorithm for Regularization of Matching Point Cloud

This paper propose a regularization method of RANSC based on the mean model to solve the big density and uneven distri-bution problem of matching point cloud.Grid as the basic unit, we use RANSAC method to choose multiple random samples repeated-ly.At the same time, the mean elevation model and established conditions are adopted to get the final agreement sets.Then we regu-larize the grid through the mean elevation model and the adjacent principle based this sets.Compared with the real data, the results show that our method can guarantee the accuracy of the data, while greatly improve the reliability and simplification of the data.