Noise-adjusted subspace linear discriminant analysis for hyperspectral-image classification

The traditional solution to addressing the small-sample-size problem as it applies to linear discriminant analysis is to implement the latter in a principal-component subspace, a strategy known as subspace linear discriminant analysis. In this work, this approach is extended by coupling subspace linear discriminant analysis and noise-adjusted principal component analysis in order to provide noise-robust feature extraction and classification of high-dimensional data. The resulting noise-adjusted subspace linear discriminant analysis is evaluated using hyperspectral imagery, with experimental results demonstrating that the proposed approach provides not only superior classification performance as compared to traditional subspace-based linear-discriminant methods but also effective dimensionality reduction for classification even in the presence of noise.