Spectral discrimination based on the optimal informative parts of the spectrum

摘要: Developments in sensor technology boost the information content of imagery collected by space- and airborne hyperspectral sensors. The sensors have narrow bands close to each other that may be highly correlated, which leads data redundancy. This paper first shows a newly developed method identify most informative spectral regions spectrum with minimum dependency other, second evaluates land cover class separability on given scenes using constructed bands. selects defined accuracy. It is applied images over three different types including vegetation, water bare soil. gives band combinations for showing regions; then discrimination analysis available classes scene carried out. Different measures based distribution scatter matrices were calculated. results show produced are well-separated classes.