Face Recognition Based on Feature Selection Strategy
Hengliang Tang · Journal of Information and Computational Science · 2014
In order to extract the efiective and discriminant face features, a useful feature selection strategy is designed for face representation. The strategy can remove the facial redundancies, extract the discriminant features, promote the calculation e‐ciency, and also guarantee the feasibility of the recognition framework. Based on this, the sparse representation framework is improved to collect all the face features and match the recognition task. The experiments, tested on ORL, YALE, CMU-PIE and FERET face databases, demonstrate that the proposed method is efiective and robust to facial pose, expression and illumination conditions to some extent.