An orthogonal least-square-based method for DEM generalization |
| |
| Authors: | C Chen Y Li |
| |
| Institution: | 1. Geomatics College , Shandong University of Science and Technology , Qingdao , Shandong , China;2. Key Laboratory of Surveying and Mapping Technology on Island and Reef , National Administration of Surveying, Mapping and Geoinformation , Qingdao , Shandong , China chencf@lreis.ac.cn;4. Department of Astronomy , Beijing Normal University , Beijing , China |
| |
| Abstract: | A new method based on orthogonal least square (OLS) of multiquadric algorithm (MOLS) is proposed for digital elevation model (DEM) generalization by the retrieval of critical points from a grid-based DEM to construct a triangulated irregular network surface. The focus is on the method for accurately obtaining the critical points, which maximally retain the important terrain feature lines. The grid-based DEM to be generalized is first approximated in terms of multiquadric method, and then the significances of the grids are assessed based on OLS method with the merit that each selected grid point gives the maximal increment to the explained variance of the desired output. We used six study sites with different terrain complexities to comparatively analyze the generalization accuracies of MOLS and the classical methods including very important method and point-additive method (Latticetin) under different numbers of retrieved significant points. The results indicate that MOLS averagely performs better than the classical methods for the original DEM generalization in terms of root mean square error. The analytical results also show that MOLS has the better ability in maintaining the feature lines inherent in the original DEM than the classical methods in terms of streamline matching rate. The perfect performance of the newly proposed method can be attributed to the high accuracy of multiquadric method for surface approximation and the effectiveness of OLS for point significance assessment. |
| |
| Keywords: | DEM generalization multiquadric orthogonal least square |
|
|
