A Study of Remote Sensing Image Landform Frame and Lithologic Component Decomposing Algorithm and Multifractal Feature of Rock Types

Landform frame-lithologic component decomposing algorithm for remote sensing (RS) image is proposed according to the principle of optical imaging and fractal feature of landform. The algorithm is applied to decompose the enhanced thematic mapper (ETM) image of rocks for the study of α-f(α) multifractal spectra. The original ETM image, decomposed landform frame and lithologic component subimage are used to calculate the α-f(α) multifractal spectra of adamellite and metamorphic sedimentary rocks in different areas. From the original ETM image, the α-f(α) multifractal spectra do not show any relation to rock types and landform. However, using the decomposed subimage, the adamellite from different areas shows the same α-f(α) spectra feature of the lithologic component subimage, but different α-f(α) spectra features of the landform frame subimage. On the contrary, the adamellite and metamorphic sedimentary rocks in the nearby area have different α-f(α) spectra of the lithologic component subimage, but rather similar α-f(α) spectra of the land frame subimage. This means the land frame-lithologic component algorithm and α-f(α) multifratal spectra can provide a new validated method to improve lithology recognition with RS image texture.