Enhanced locality sensitive discriminant analysis for image recognition
Jiwen Lu · Electronics Letters · 2010
An improved manifold learning method, called enhanced locality sensitive discriminant analysis (ELSDA), for image recognition is proposed. Motivated by the fact that statistically uncorrelated and parameter-free are two desirable and promising characteristics for feature extraction, a new difference-based optimisation objective function with uncorrelated constraint for appearance-based image recognition has been designed. Experimental results demonstrate the efficacy of the proposed method.