Hyperspectral Remote Sensing Rock and Mineral Spectral Feature Mining Based on Rough Set Theory

The hyperspectral RS is a new technology for fine spectral feature, but the spectral feature of same class object are still not completely consistent, because the ground surface environment is complex. In this paper, we analyze the spectral characteristics and its fractal characteristics, and construct a spectrum curve feature matrix, which consist of center distance, informational entropy, fractal dimension, peak value, valley value, wavelength of the wave crests and two wave troughs. Then we apply the approximate classing of rough set theory to distinguish the rock and mineral from Hyperspectral image.