Abstract：Due to the restriction of spectral and spatial resolution, it is difficult to meet the requirements of extracting land-cover elements accurately by using single remote sensing data to distinguish and classify land types.This paper used CBERS-02B satellite panchromatic band data and multi-spectral CCD data as information sources for making full use of the different information and complementing the shortage of each other.Then, pixel-based PCA and principal component transformation method were used to process the information together.On this basis, the patterns of land cover elements in mining areas disturbed could be extracted and classified.The experiment achieved good results.

Keyword：multi-source remote sensing;CBERS-02B Satellite;fusion;land cover;classification;