Itis provedthatsome effectshave come outforrockinformation separationfrom remotesensing data usingthe principalcomponentanalysis.New information

relatively weak

isrevealed bytwo

c

omponentscatter

p

lotclassification

butit does notoccurin the centerarea ofthe aggregation .These phenomenaindicate thatthere are some other relative weak featuresin multi

b

and remote sensing raw data

beside main features presented by the principal componentanaly sis.The principalcomponentanalysisis onlyone resultoflineartransformation .In orderto derive more potentialinforma tion from multi

b

and raw data ofremote sensing

the step

w

ise orthogonaltransformation software is developed to extract potentialinformation ofrocktypes. Meanwhile

the maskingtechniqueis usedto getrid oftheinterference of water body and dry riverfrom theraw data.Goodresultsare achieved andthe methodisalso usedin other placesofthe Huabeiplat form rim .