Context Dependent Spectral Unmixing

Abstract

A hyperspectral unmixing algorithm that finds multiple sets of endmembers is introduced. The algorithm, called Context Dependent Spectral Unmixing (CDSU), is a local approach that adapts the unmixing to different regions of the spectral space. It is based on a novel objective function that combines context identification and unmixing into a joint function. This objective function models contexts as compact clusters and uses the linear mixing model as the basis for unmixing. The unmixing provides optimal endmembers and abundances for each context. An alternating optimization algorithm is derived. The performance of the CDSU algorithm is evaluated using synthetic and real data. We show that the proposed method can identify meaningful and coherent contexts, and appropriate endmembers within each context.