LZD模型多维粗差发现和定位能力分析

Detectable and locatable ability of multiple gross errors in LZD matching algorithm

不借助地面控制点的DEM匹配差异探测算法是多时相DEM分析的最新发展方向。现有的算法大多把DEM表面差异作为粗差进行处理,算法性能得到了较大的提高,但忽略了对DEM匹配模型多维粗差的可发现和可定位能力进行分析,而这是DEM差异探测算法的理论基础。本文采用测量误差和可靠性理论中的判断矩阵法,通过理论分析和模拟试验验证了对于实际DEM而言,LZD模型具备多维粗差发现和定位能力。基于LZD模型的DEM差异探测算法能够正确地探测表面变形,这就完善了DEM差异探测研究的理论基础。

DEM matching for deformation detection without ground control points is a hot research topic in the field ot the multi-temporal DEM analyses. Ahnost all existing methods take the surface deformation as gross errors, and improve the performance greatly. However, analyzing of detectable and locatable ability of multiple gross errors in LZD matching algorithm is neglected. The detectable and locatable ability of gross error is the precondition in theory for these methods. This paper adopts the judgement matrix method in surveying error and reliability theory to study this problem. The results of the theoretic and experimental analyses show that the observations in LZD algorithm model own the detectable and loeatable ability of muhiple gross errors. Therefore, the DEM matching for deformation detection can correctly detect the surface deformation, and its foundation in theory is perfect.