Parallel unmixing of hyperspectral data using complexity pursuit

Stefan A. Robila, Martin Butler

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Accurate and fast data unmixing is key to most applications employing hyperspectral data. Among the large number unmixing approaches, Blind Source Separation (BSS) has been employed successfully through a variety of techniques, yet most of these approaches continue to be computationally expensive due to their iterative nature. In this context, it is imperative to seek efficient approaches that leverage the accuracy of the algorithms and the availability of off-the-shelf computationally performant systems such as multi-cpu and multi core. In this paper we tackle the spatial complexity based unmixing, a new technique shown to outperform many BSS solutions. We develop a new parallel algorithm that, without decreasing the accuracy ensures significant computational speedup when compared to the original technique. We provide a theoretical analysis on its equivalency with the algorithm. Furthermore we show through both complexity analysis and experimental results that the algorithm provides a speedup in execution linear to the number of computing cores used.

Original languageEnglish
Title of host publication2010 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2010
Pages1035-1038
Number of pages4
DOIs
StatePublished - 1 Dec 2010
Event2010 30th IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2010 - Honolulu, HI, United States
Duration: 25 Jul 201030 Jul 2010

Publication series

NameInternational Geoscience and Remote Sensing Symposium (IGARSS)

Other

Other2010 30th IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2010
CountryUnited States
CityHonolulu, HI
Period25/07/1030/07/10

Keywords

  • Blind source separation
  • Complexity pursuit
  • High performance computing
  • Hyperspectral imagery
  • Linear unmixing

Fingerprint Dive into the research topics of 'Parallel unmixing of hyperspectral data using complexity pursuit'. Together they form a unique fingerprint.

  • Cite this

    Robila, S. A., & Butler, M. (2010). Parallel unmixing of hyperspectral data using complexity pursuit. In 2010 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2010 (pp. 1035-1038). [5648919] (International Geoscience and Remote Sensing Symposium (IGARSS)). https://doi.org/10.1109/IGARSS.2010.5648919