A fuzzy spectral clustering for lossless and multiresolution coding of hyperspectral images

Abstract

In this paper, we propose a two-stage algorithm for hyperspectral image coding in a multiresolution way. The first stage consists in classifying the spectral components into separate clusters sharing similar characteristics. For this purpose, we have applied a fuzzy clustering algorithm which allows the clusters to have a flexible number of components. In the second stage, a vector lifting scheme is applied within each cluster in order to exploit simultaneously the spatial and spectral redundancies. A selective process is also applied to switch between inter-component and intracomponent coding of a given component. Experiments carried out on AVIRIS images indicate the outperformance of the proposed method w.r.t. the state-of-art coders.