engleski

A Comparative Study of PCA, ICA AND LDA

Different statistical methods for face recognition have been proposed in recent years and different research groups have reported contradictory results when comparing them. The goal of this paper is to present an independent, comparative study of three most popular appearance-based face recognition algorithms (PCA, ICA and LDA) in completely equal working conditions. The motivation was the lack of direct and detailed independent comparisons in all possible algorithm implementations (e.g. all algorithm-metric combinations). FERET data set will be used for consistency with other studies. It will be shown that no particular algorithm-metric combination is the optimal across all standard FERET tests and that choice of appropriate algorithm-metric combination can only be made for a specific task.

face recognition; PCA; ICA; LDA; FERET; subspace analysis methods

nije evidentirano