Hyperspectral imagery visualization using band selection

This paper investigates hyperspectral image display based on selection of three spectral channels to build a red-green-blue (RGB) composite. A series of band selection algorithms are implemented and compared for this purpose. In particular, three color composition schemes based on visualization-oriented spectral segmentations are proposed. The resulting color displays are evaluated in terms of class separability using a statistical classifier and perceptual color distance. Results demonstrate that three-band selection is a simple and effective technique for hyprespectral image visualization.