Weight Optimized Fusion in Frequency Domain of Multi-DEM by Wavelet Analysis
ISBN 978-85-88783-11-9
Authors
1Liu, P.; 2Sun, L.; 3Yan, W.; 4Hou, S.; 5Zhang, W.
1XI’AN RESEARCH INSTITUTE OF SURVEYING AND MAPPING Email: 24732823@qq.com
293995 TROOPS Email: sunliang1218@163.com
3XI'AN MAPPING CENTER Email: 1150442793@qq.com
4XI’AN RESEARCH INSTITUTE OF SURVEYING AND MAPPING Email: 24732823@qq.com
5XI’AN RESEARCH INSTITUTE OF SURVEYING AND MAPPING Email: 24732823@qq.com
Abstract
In order to utilize the complementary information, multi-DEM fusion in frequency domain by wavelet multiresolution analysis is presented. Two arbitrary DEMS are preprocessed via re-sampled and co-registered and systemic error eliminated, then the low frequency coefficient and high frequency coefficient are obtained recur to wavelet decomposition, different weight are assigned to the low frequency coefficient and high frequency coefficient, the performance index of RMSE is optimized through ergodic process of weight, the fused DEM is acquired after wavelet inverse transformation. Typical physiognomy character of experiment area covering mountain and valley is selected, corresponding DEM data of SRTM DEM(Shuttle Radar Topography Mission DEM) and ASTER DEM(Advanced Space borne Thermal Emission and Reflection Radiometer Digital Elevation Mode) is prepared. The precision of data before and after fusion are compared with reference DEM data respectively, collectivity statistic and spot check show that the precision of fusion DEM has improved relative to the original DEM, weight optimized fusion provides an effective and reliable method for DEM product from multi-DEM.
Keywords
Weight optimized; Ergodic Process of Weight; Wavelet Multiresolution Analys