L_p估计的计算问题

The Computational Problem for L_p Estimation

本文对Ｌｐ估计的直接算法进行了详细的分析和讨论，由于直接算法效率低，文中提出了一种快速算法（残差逐步下降法），采用可变尺度因子来修正Ｈｕｂｅｒ的常数尺度因子。最后，以实例加以分析和说明。

The direct algorithm of Lp estimation is analysed and discussed in detail. A fast algorithm (residual steepest descent) which can revise Huber's constant scale factor using a variable scale factor is proposed because the computational efficiency of direct algorithm is very low. Finally. the analysis and explanation is given by numerical example.