一种改进的无控制DEM匹配算法

An improved algorithm of DEM matching without control

DEM配准是数据融合和超分辨率DEM重建的一个重要步骤,在一些地形条件极其复杂的地区,观测受到限制,很难在图上选取同名控制点进行匹配。在无控制DEM匹配中,ICP算法迭代收敛速度很快,但因计算距离复杂,整体计算效率不高,且对于存在尺度变换的数据无法正确匹配;LZD算法迭代收敛速度较慢,但计算量小,计算效率很高。针对两种经典无控制匹配算法存在的问题,提出一种融合ICP和LZD算法的改进无控制DEM匹配算法,并在最小二乘估计的基础上引入差分模型求模型的抗差解,削弱由于地表变形对匹配结果的影响。最后,通过实验对本文算法的可行性和可靠性进行了分析和验证。

DEM registration is an important step in data fusion and super resolution DEM reconstruction. In some areas with extremely complicated terrain conditions, it is difficult to match the same control points on the map due to the limitation of observation. In the DEM matching without control, the convergence speed of ICP algorithm is very fast, but because of the distance computation is more complex, the overall computational efficiency is not high, and for the existence of scale transformation of the data can not be correctly matched. LZD iterative algorithm converges slowly, but a small amount of calculation, high calculation efficiency. According to the two classic matching algorithm without control problems, it puts forward an improved LZD algorithm combining ICP and DEM matching without control algorithm, and introduces the difference model for robust solution model based on least squares estimation, weakened due to the impact of the matching results of surface deformation. Finally, the feasibility and reliability of the proposed algorithm are analyzed and verified by experiments.