Detection And Validation Of The Variation In Digital Elevation Model

The source of the variation of digital elevation model data mainly comes from subjective data error(intentional replace the data in some vital area) and objective data error(gross error that exists in the process of measurement).Data variation has great impact when it is applied into practical application. The purpose of this research is to describe the variation among different DEMs in order to provide the basis for erroneous finding in DEM.ASTER GDEM(Advanced Spaceborne Thermal Emission and Reflection Radiometer Globe Digital Elevation Model)is the most complete global digital elevation data so far and is free accessible to users all over the world, real time is one of its another advantage.Currently, ASTER GDEM has already became one of the most important data resource in scientific research and application fields.This paper uses ASTER GDEM as test data,and regards the DLR DEM which provided by German Aerospace center as reference contrastive data. There are two detailed research methods that I applied in this research. In the first method, I proposed a matching method for remote sensing images, its combined the SURF(speeded up robust features) algorithm and RANSAC(random sample consensus)algorithm to achieve feature point matching. This method not only inherited the speed of SURF algorithm, but also assured the precision of image matching. After that, using the x and y coordinate of the matching points to compute their horizontal displacement and draw the 3D map of the calculated horizontal displacement of the ASTER GDEM by contrasting with DLR DEM. The second method applied professional GIS software ArcGis to fulfil the extraction and optimization of the feature points in small area, then matching those points with the assistance of the contour and relational position of the feature points.Plotting the horizontal displacement of ASTER GDEM by comparing it with DLR DEM with the assist of matlab and comparing those two results from different method in the same experiment area. The results show that:(1) The RANSAC algorithm can solve the mismatching problem in the process of image matching. The method of combination of SURF algorithm and RANSAC algorithm can achieve accurate result in image matching.(2)By contrasting the horizontal displacement of ASTER GDEM relative to DLR DEM from those two different research methods, we can draw the conclusion that the results from those two approach have similar effect when focus on horizontal displacement with distance from 0-30 m and 30-60 m. The amount of the feature points is fewer and maximum horizontal displacement distance is larger when applied the method algorithm based on image matching.It is much accuracy to match those feature points by hand in the ArcGis,however, the workload of manual point matching is too heavy to fulfil that on large area.