Research Papers

Comparison of hyperspectral endmember extraction algorithms

[+] Author Affiliations
Jee-cheng Wu

National I-Lan University, Department of Civil Engineering, No. 1, Sec. 1, Sheng-Lung Road, I-Lan City 260, Taiwan

Gwo-chyang Tsuei

National I-Lan University, Department of Civil Engineering, No. 1, Sec. 1, Sheng-Lung Road, I-Lan City 260, Taiwan

J. Appl. Remote Sens. 7(1), 073525 (Aug 09, 2013). doi:10.1117/1.JRS.7.073525
History: Received March 8, 2013; Revised July 3, 2013; Accepted July 11, 2013
Text Size: A A A

Abstract.  In hyperspectral imagery, endmember extraction is the process of finding a pure spectrum set within the materials present in a hyperspectral scene. However, various endmember extraction algorithms (EEAs) can yield different endmember spectrum sets. This research presents a comparison of four EEAs: pixel purity index, automatic target generation process (ATGP), N-finder (N-FINDR), and simplex identification via split augmented Lagrangian. To perform the comparison, a ground reference geological map is first coregistered with the hyperspectral scene. Then, 12 geological ground truth spectra are chosen. The four EEAs are then used to extract endmember spectra from the scene. Next, the extracted endmember spectra are applied to generate abundance maps using fully constrained least squares. The largest spectrum magnitude in the abundance map is considered the endmember. Finally, the spectral angle mapper and root-mean-square error between the extracted endmember spectrum and the chosen geological spectral spectra are computed, using the angle and minimum error to evaluate similarities. The results of this EEA comparison show that only the ATGP algorithm could consistently identify endmembers with the least computation time. Additionally, the N-FINDR algorithm is able to extract the most endmembers with the closest endmember similarity measures, although this required the highest computation time.

Figures in this Article
© 2013 Society of Photo-Optical Instrumentation Engineers

Topics

Algorithms ; Minerals

Citation

Jee-cheng Wu and Gwo-chyang Tsuei
"Comparison of hyperspectral endmember extraction algorithms", J. Appl. Remote Sens. 7(1), 073525 (Aug 09, 2013). ; http://dx.doi.org/10.1117/1.JRS.7.073525


Access This Article
Sign in or Create a personal account to Buy this article ($20 for members, $25 for non-members).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Related Book Chapters

Topic Collections

PubMed Articles
Advertisement


 

  • Don't have an account?
  • Subscribe to the SPIE Digital Library
  • Create a FREE account to sign up for Digital Library content alerts and gain access to institutional subscriptions remotely.
Access This Article
Sign in or Create a personal account to Buy this article ($20 for members, $25 for non-members).
Access This Proceeding
Sign in or Create a personal account to Buy this article ($15 for members, $18 for non-members).
Access This Chapter

Access to SPIE eBooks is limited to subscribing institutions and is not available as part of a personal subscription. Print or electronic versions of individual SPIE books may be purchased via SPIE.org.