稳健初值的选权迭代法剔除DEM数据粗差

Gross error elimination in DEM data by weight iteration with initial values

本文将抗差估计的思想融入到数字高程模型粗差探测的算法中,设计出对模型误差、特别是粗差具有抵抗能力的粗差探测算法。稳健初值的选权迭代法即为抗差估计的一种,首先是通过线性规划的单纯形解法来计算观测值的残差,然后再根据残差和有关的参数,按所选择的权函数,计算每个观测值的权,经过迭代计算求得观测值的残差,然后按照统计检验的方法剔除粗差。通过实验证明,具有稳健初值的选权迭代法具有很强的稳健性和粗差探测能力。

This article merged the idea of robust estimation into the algorithm of detecting gross errors in DEM,in order to design the algorithm of resisting the model errors,especially gross errors.Robust Estimation contains the method of weight iteration with initial values.First,it was by linear programming method to calculate the residual errors of observed values.Second,based on the residual errors and parameters,the new weight value for each observation according to the weight function was figured up.Third,the residual errors of the observation was obtained through iterative calculation.Then the statistical methods were used to eliminate gross errors.Finally,the experiment showed that the studied method had strong robustness and good ability to detecting gross errors.