17 May 2006 Automatic clustering based on an information-theoretic approach with application to spectral anomaly detection
Author Affiliations +
Abstract
An information-theoretic method is described for automatically determining the best number of clusters. It is motivated by Rissanen's minimum description length principle that states the best representation is the one with the fewest bits. The method is evaluated using two different clustering algorithms: a mode finder based on scale-space algorithm, and a vector quantizer (VQ). Synthetic, single- and multi-band image clustering examples are presented. Clusterings produced by the mode finder are shown to better correspond to distinguishable surface categories in the scene than those produced by the VQ algorithm. VQ clusterings are evaluated within an anomaly detector, which detects manmade object/changes as spectral outliers within a set of background clusters. It is shown that the optimal VQ clustering (the one with the fewest bits) produces the best detection performance.
© (2006) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mark J. Carlotto, Mark J. Carlotto, } "Automatic clustering based on an information-theoretic approach with application to spectral anomaly detection", Proc. SPIE 6235, Signal Processing, Sensor Fusion, and Target Recognition XV, 623517 (17 May 2006); doi: 10.1117/12.668805; https://doi.org/10.1117/12.668805
PROCEEDINGS
9 PAGES


SHARE
RELATED CONTENT

Modeling hyperspectral imaging data
Proceedings of SPIE (September 23 2003)
Multiple frame cluster tracking
Proceedings of SPIE (August 07 2002)
Statistics of hyperspectral imaging data
Proceedings of SPIE (August 20 2001)
Group assignment problem
Proceedings of SPIE (January 05 2004)

Back to Top