两期变形网L_2/L_1范数解法

A Solution of L_2/L_1 Norm for Deformation Network in two Terms

本文将L_2范数和L_1范数结合起来,利用L_1范数的稳健性,提出了用于变形网的两阶段L_2/L_1范数解法,并通过算例将它与最小二乘法进行比较,其结果表明网点中有单点位移存在时最小二乘方法会得到有偏估计,而L_2/L_1范数解法能可靠地得出网点中各点的变动信息及位移值。

According to the robust of L_1 norm, this paper combines L_2 norm with L_1 norm and gives a solution of L_2/L_1 norm in two phases for deformation network. Comparing with least squares method by using two examples, the results are as follows: if there is a displacement of single point in the network, the estimate of least squares method is biased, but the solution of L_2/L_1 norm can reliably obtain the changed information and displacement of each point.