Hyperspectral Subspace Identification

• Published in: IEEE transactions on geoscience and remote sensing Vol. 46; no. 8; pp. 2435 - 2445
• Main Authors: Bioucas-Dias, J.M., Nascimento, J.M.P.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.08.2008; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Applied geophysics; Change detection algorithms; Data storage; Dimensionality reduction; Earth sciences; Earth, ocean, space; Eigenvalues; Eigenvalues and eigenfunctions; Exact sciences and technology; hyperspectral imagery; Hyperspectral imaging; Hyperspectral sensors; hyperspectral signal subspace identification by minimum error (HySime); hyperspectral unmixing; Internal geophysics; Least mean squares algorithm; linear mixture; Matrices; Matrix decomposition; Memory; minimum mean square error (mse); Object detection; Parameter estimation; Performance gain; Signal processing; subspace identification; Subspaces; Tuning; Dimensionality reduction; linear mixture; minimum mean square error (mse); hyperspectral imagery; subspace identification; hyperspectral unmixing; hyperspectral signal subspace identification by minimum error (HySime); algorithms; eigenvalues; remote sensing; decomposition; classification; data storage; correlation; algorithm performance; imagery; noise; reduction; errors
• ISSN: 0196-2892; 1558-0644
• DOI: 10.1109/TGRS.2008.918089

Signal subspace identification is a crucial first step in many hyperspectral processing algorithms such as target detection, change detection, classification, and unmixing. The identification of this subspace enables a correct dimensionality reduction, yielding gains in algorithm performance and complexity and in data storage. This paper introduces a new minimum mean square error-based approach to infer the signal subspace in hyperspectral imagery. The method, which is termed hyperspectral signal identification by minimum error, is eigen decomposition based, unsupervised, and fully automatic (i.e., it does not depend on any tuning parameters). It first estimates the signal and noise correlation matrices and then selects the subset of eigenvalues that best represents the signal subspace in the least squared error sense. State-of-the-art performance of the proposed method is illustrated by using simulated and real hyperspectral images.