Multi-scale Generalization Of DEM Based On Wavelet Transform

With constant enlargement of the application domain of GIS and the increasing need for GIS, the approach of applying multi-scale spatial data to the observation and analysis of geographic phenomena from different perspectives becomes more and more important. Multi-scale processing and representation of spatial data has become the crucial element of building multi-scale spatial database and accomplishing multi-scale GIS or Scale-free GIS. And the establishment of the scale-dependent spatial data representation model and the realization of the auto-generalization of geographic information is the key technology of resolving multi-scale processing and representation of spatial data.Multi-Resolution Analysis (MRA) notion of wavelet theory has striking similarity with multi-scale representation of spatial data. In this thesis, based on the deep analysis of the essence of the representation of multi-scale spatial data, the author puts forward the approach of using wavelet multi-resolution analysis theory to analyze the geographic space from multiplied scale. The multi-scale auto-generalization and the consecutive representation of Digital Elevation Model (DEM) can be achieved by building and maintaining dynamically a subsequence of multi-scale nested geographic space and the scale-dependent representation model of spatial data.Therefore, the notion of wavelet transform theory and the MRA are illustrated in detail at the beginning of this thesis. Then, the thesis further explores the expediency of wavelet for multi-scale generalization DEM, the processing way of DEM boundaries and the selected standard of threshold for the wavelet decomposed high frequency coefficient, which supply a necessary preparation for DEM muti-scale generalization. By using binary system , multi-band and lifting wavelet, the author has done comprehensive experiments on multi-scale generalization for different typeof terrain DEM , and has studied the impact of different wavelet transform way for generalized DEM on elevation accuracy and spatial correlation. At last, the thesis illustrates the impact of wavelet multi-scale generalization of DEM on terrain factor and its changing rule.The experiments prove that, for multi-scale generalization of DEM, it is easy to derive multi-scale data frames and accomplish consecutive representation by using wavelet transform. And the impact of generalization processing on elevation accuracy can be completely controlled in the error range of corresponding scale DEM. At the same time, the influences of generalized scale on the factors of basically characteristic terrain take distinct regular.