Ensemble of Target Detection Methods on Hyperspectral Imagery

Creative Commons License


1st INTERNATIONAL CONFERENCE ON ADVANCES IN SCIENCE (ICAS 2016), İstanbul, Turkey, 31 August 2016, pp.146-149

  • Publication Type: Conference Paper / Full Text
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.146-149


Target detection is a popular research topic in high
dimensional data processing area. A great number of detection
algorithms have been developed and used in the past three
decades, such as: Adaptive Cosine Estimator (ACE),
Constrained Energy Minimization (CEM), Spectral Angle
Mapper (SAM), Covariance Descriptor (CD) and Matched Filter
(MF). Every method has several advantageous and
disadvantageous in their working principle. In this paper, we
propose ensemble based learning techniques to increase the
detection algorithms efficiency. In machine learning, an
ensemble approach is one of the promising approaches to acquire
better performance than any other learning algorithms. In this
paper, we combine aforementioned detection algorithms via
ensemble techniques. To verify the performance improvement, two
high dimensional data are processed for detection purpose. The
obtained results show that the ensemble based technique produce
better performance than the conventional detection algorithms.