Chapter 8. One-Class SVMs for Hyperspectral Anomaly Detection

  1. Dr Gustavo Camps-Valls B.Sc., Ph.D. professor member2 and
  2. Dr Lorenzo Bruzzone M.S., Ph.D. Postdoctoral Researcher Professor member Chair3
  1. Amit Banerjee,
  2. Philippe Burlina and
  3. Chris Diehl

Published Online: 4 NOV 2009

DOI: 10.1002/9780470748992.ch8

Kernel Methods for Remote Sensing Data Analysis

Kernel Methods for Remote Sensing Data Analysis

How to Cite

Banerjee, A., Burlina, P. and Diehl, C. (2009) One-Class SVMs for Hyperspectral Anomaly Detection, in Kernel Methods for Remote Sensing Data Analysis (eds G. Camps-Valls and L. Bruzzone), John Wiley & Sons, Ltd, Chichester, UK. doi: 10.1002/9780470748992.ch8

Editor Information

  1. 2

    Image Processing Laboratory (IPL) & Dept. Enginyeria Electrónica, Universitat de Valéncia, Spain

  2. 3

    Dept. Information Engineering and Computer Science, University of Trento, Italy

Author Information

  1. Applied Physics Laboratory, The Johns Hopkins University, USA

Publication History

  1. Published Online: 4 NOV 2009
  2. Published Print: 23 OCT 2009

ISBN Information

Print ISBN: 9780470722114

Online ISBN: 9780470748992

SEARCH

Keywords:

  • one-class SVMs for hyperspectral anomaly detection;
  • support vector framework for hyperspectral anomaly detection;
  • Support Vector Data Description (SVDD);
  • goodness-of-fit test statistic for hyperspectral imagery based on Barringhaus, Henze, Epps and Pully (BHEP) test;
  • SVDD derivation;
  • SVDD function optimization;
  • SVDD algorithms for hyperspectral anomaly detection;
  • computational and detection performance of SVDD anomaly detectors;
  • normalized metric appropriate for anomaly detection in spectral imagery

Summary

This chapter contains sections titled:

  • Introduction

  • Deriving the SVDD

  • SVDD function optimization

  • SVDD algorithms for hyperspectral anomaly detection

  • Experimental results

  • Conclusions

  • References