采用截尾最小二乘估计的DEM匹配方法

DEM Matching Algorithm Using Least Trimmed Squares Estimator

不借助控制点匹配表面来探测表面差异是一个非常困难的问题,在多时相DEM分析中具有非常广阔的应用前景。提出一个使用截尾最小二乘估计的最小高差算法(LTS-LZD),该算法在迭代过程中通过基于高差直方图的自适应阈值来区分变形区观测量。通过模拟试验对算法性能进行全面深入的试验,试验结果表明新方法具有较高的匹配精度与差异探测精度,且与变形比例基本无关。

The surface matching for change detection without control points is inherently very difficult problem, and has potential application for multi-temporal DEM analyses. The Least Z-Difference （LZD） algorithm using Least Trimmed Squares estimator （LTS-LZD） is proposed in this paper. The novel algorithm employs a self-adaptive threshold based on the histogram of height difference to identify observations located in deformed region. The simulated experiments are adopted to test the new algorithm, and the results show it has higher matching and deformation detecting accuracy, and is less affected by the percentage of changes,