SAM weighted KEST algorithm for anomaly detection in hyperspectral imagery
DOI:
Author:
Affiliation:

Jiangsu Key Laboratory of Spectral Imaging and Intelligent Sence, Nanjing University of Science and Technology,Jiangsu Key Laboratory of Spectral Imaging and Intelligent Sence, Nanjing University of Science and Technology,Jiangsu Key Laboratory of Spectral Imaging and Intelligent Sence, Nanjing University of Science and Technology,Jiangsu Key Laboratory of Spectral Imaging and Intelligent Sence, Nanjing University of Science and Technology,Jiangsu Key Laboratory of Spectral Imaging and Intelligent Sence, Nanjing University of Science and Technology

Clc Number:

Fund Project:

  • Article
  • |
  • Figures
  • |
  • Metrics
  • |
  • Reference
  • |
  • Related
  • |
  • Cited by
  • |
  • Materials
  • |
  • Comments
    Abstract:

    A SAM weighted KEST algorithm based on kernel eigenspace separation transform (KEST) was proposed for anomaly detection in hyperspectral imaging. Weights are introduced for each sample in the difference correlation matrix (DCOR), and the input pixel neighbor surroundings. All samples were weighted according to the angle between the sample spectral vector and the centered vector in detection window to minimize the influence of anomalous data and outstand the contribution of principle component. In this way, DCOR represented the difference between target and background distribution much better. Experimental results indicate that the proposed method shows superior performance over the conventional anomaly detection algorithms and KEST.

    Reference
    Related
    Cited by
Get Citation

HAN Jing, YUE Jiang, ZHANG Yi, BAI Lian-fa, CHEN Qian. SAM weighted KEST algorithm for anomaly detection in hyperspectral imagery[J]. Journal of Infrared and Millimeter Waves,2013,32(4):359~365

Copy
Share
Article Metrics
  • Abstract:
  • PDF:
  • HTML:
  • Cited by:
History
  • Received:February 27,2012
  • Revised:October 29,2012
  • Adopted:August 23,2012
  • Online: August 29,2013
  • Published: