Unmixing Analysis of Close-Range Hyperspectral Images

This paper presents the unmixing analysis of hyperspectral images collected at close-range of vegetation plots associated with AON-ITEX in Utqiaġvik, AK. The resulting analysis illustrates the potential value of unmixing for the analysis of very high spatial resolution hyperspectral images. Unmixing can extract the signatures of the materials present in the image and capture their spectral variability and spatial distribution to support further analysis up the image chain. A variation of PPI is presented that is suitable for this application and address the problem of estimating the number of endmembers from the data.