DEM融合算法及其在困难地区测图中的应用

对于我国西部高山区,如横断山脉,高程起伏明显,常年被云雾覆盖,日照稀少,采用传统方法进行地形图测绘存在较大困难,依赖单一方法获取的DEM往往难以满足测图的精度要求。为充分利用不同传感器和不同方法生成的DEM的优点,本文根据各方法的特点,结合小比例尺地形图中低精度的DEM,基于绝对精度等先验知识确定优先级别、相关/干系数确定融合权重,提出了一种包括雷达干涉测量、光学立体摄影测量、不同侧视方向像对雷达立体测图生成的四种多源DEM的像素级融合算法。在横断山脉地区使用所提融合算法进行了实验,获得了一个总体精度得到提高的无缝DEM,实验结果表明新算法为地形复杂的测图困难地区DEM获取提供了一种可能的解决方案。

DEM Fusion and Its Application to Topography Mapping in Complex Areas

It is difficult to map the topography of high hilly areas in the west of China using traditional methods .For example Hengdu-an Mountain area, where is covered by the clouds, fog and mist,cannot be shined by sunlight most days all over the year, with eleva-tions of very large range.It is very challenging to produce the DEM accurately depending on a single method.This paper aims at utili-zing the advantages of several DEMs generated using different methods or from images collected by different sensors, to provide a gap-less DEM and improve the overall accuracy.Based on the characteristics of DEMs generated by InSAR, optical photogrammetry, rada-rgrammetry from opposite directions, the four DEMs are fused taking each one’ s advantages, supported by a low resolution Dem in to-pography maps with a smaller scale.The DEMs are processed with specific preferences defined by their absolute accuracy and weights defined by their correlation or coherence coefficient using a pixel-based methods.A gapless DEM with an enhanced accuracy was ob-tained using the proposed fusion method.In conclusion, the new method is a potential solution to produce DEM in mapping topography of complex areas.Results of Hengduan test site are analysed and show an improvement of the fused DEM.