Independent Innovation Analysis for Hyperspectral Imagery Unmixing

Abstract

Hyperspectral imagery (HSI) unmixing can be seen as a blind source separation (BSS) process, thus, independent component analysis (ICA) has recently been used widely as a useful tool to unmixing hyperspectral data. It models a mixed pixel as a linear mixture of the constituent (end member) spectra weighted by the correspondent abundance fractions. However, the unmixing results of ICA are not satisfied because ICA demands the sources are statistically independent, but usually, the sources are not statistically independent for the real hyperspectral data. In this paper, a BSS algorithm called independent innovation analysis (IIA) is introduced. The proposed algorithm is based on the mutual independency of the innovations of source signals instead of original signals. This algorithm takes into account both the temporal structure and the high-order statistics of source signals and in contrast to the most known blind separation or ICA algorithms only exploiting the second order statistics or the non-Gaussianity. The hyperspectral imagery unmixing experimental results show that IIA provides a promising approach to unmix HSI.