Image visualisation and classification of MODIS/ASTER Airborne Simulator (MASTER) remotely sensed data for agricultural area

Abstract

A Best Band Selection Index (BBSI) algorithm to select the best band combination for image visualization and classification of high spectral resolution remotely sensed dataset was introduced in this paper. The BBSI is calculated by two components, one based on class mean (or cluster mean) difference and the other based on correlation coefficients. Using MODIS/ASTER A i r b o r n e Simulator (MASTER) images taken over Jertih, Te rengganu in 2000 as the test dataset, the BBSI correctly predicted the best t h ree-band combination that provided useful information for visualization of the image to collect training samples in superv i s e d classification. The BBSI also accurately selected the best four-band combination that produced high overall accuracy classification map with value of 89.7%.