Band Selection for Hyperspectral Imagery Using Affinity Propagation

Hyperspectral imagery generally contains enormous amounts of data due to hundreds of spectral bands. Band selection is often adopted firstly to reduce computational cost and accelerate knowledge discovery of subsequent classificationand analysis. Recently, a new clustering algorithm, named "affinity propagation," is proposed. Different from the popular k-centers clustering technique, affinity propagation operates by simultaneously considering all data points as potential cluster centers (called "exemplars") and exchanging messages between data points until a good set of exemplars and clusters emerges. In this paper, we apply affinity propagation for band selection of hyperspectral data. Experimental results demonstrate that, compared with some relevant and recent methods for band selection, the bands chosen by affinity propagation best represent the hyperspectral imagery from the pixel image classification standpoint.