Abstract. In order to analyze hyperspectral image data, pixel unmixing process is used. Pixel unmixing process involves identifying the pure pixels from the data cube. The pure pixels identified are known as “endmembers”. Endmembers are the macroscopic objects available in the scene. It can be either vegetation, water, mineral, soil, leaf of an apple tree, animal, metal or any natural or man-made material. Hyperspectral analysis includes estimation of end members available in a hyperspectral image (HSI) and their abundance in each pixel of the scene. Further, the endmembers which have been recognized can be used to plot in the spatial distribution, and interpreted by associating its signature and abundances. Research is proceeding towards finding the spectral endmembers automatically with the HSI data sets. Pixel purity index (PPI), is base of all the endmember extraction algorithms. PPI algorithm follows non iterative procedure which may end up with false endmember identification for few cases which is addressed in this paper.

Received: March 12, 2015

AMS Subject Classification:

Key Words and Phrases: endmember, hyperspectral image (HSI), pixel purity index (PPI), N-FINDR, unmixing, mixed pixel

Download paper from here.

---

DOI: 10.12732/ijpam.v101i7.21 How to cite this paper?

Source: International Journal of Pure and Applied Mathematics
ISSN printed version: 1311-8080
ISSN on-line version: 1314-3395
Year: 2015
Volume: 101
Issue: 7
Pages: 809 - 829

Google Scholar; DOI (International DOI Foundation); WorldCAT.

This work is licensed under the Creative Commons Attribution International License (CC BY).

---