Endmember Bundle Extraction Based on Multiobjective Optimization

A number of endmember extraction methods have been developed to identify pure pixels in hyperspectral images (HSIs). The majority them use only one spectrum represent kind material, which ignores the spectral variability problem that particularly characterizes a HSI with high spatial resolution. Only few algorithms multiple endmembers representing within each class, called bundle (EBE). This article introduces multiobjective particle swarm optimization for identification spectra variability. Unlike existing convex geometry-based EBE methods, operate on single geometry dataspace, proposed method divides observed data into subsets along dimension and simultaneously operates dataspaces obtain candidate based optimization. are then refined by post-processing sequential forward floating selection produce final result. Experiments conducted both synthetic real demonstrate effectiveness comparison several state-of-the-art methods.