Abstract: This study selected rich potassium rock mass and its surrounding area in Mougou,Song county,Henan province as research area,and tried to find out a method that could automatically extract out the information of the rich potassium rock from ASTER data in this area.Through various transformations of spectrum from sample points,based on multiple stepwise regression analysis technology,nine kinds of oxide content inversion model were established,including MgO,CaO,Na₂O,FeO,Fe₂O₃,K₂O,Al₂O₃,MnO,TiO₂.Using ASTER remote sensing data,inversion of the oxide content in the study area were gotten.CIPW (standard mineral calculation) algorithm was achieved based on IDL,and mass fraction of main standard minerals in the study area including quartz,albite,orthoclase and calcium feldspar were obtained.Finally,on the basis of arithmetic operations and the ratio operations,the extraction rules of the rich potassium rock were built,and the rich potassium rock mass was extracted.The rich potassium rock mass extracted in the study area was generally consistent with the known range,but the overall scope of rock mass was smaller and some new small rock mass were found.Besides,relevant analysis and discussion were made upon the extraction results.