Coupled Kernel Fisher Discriminative Analysis for Low-Resolution Face Recognition

Xiaoying Wang, Le Liu, Haifeng Hu

Abstract: In this paper, we propose a novel approach called coupled kernel fisher discriminative analysis (CKFDA) based on simultaneous discriminant analysis (SDA) for LR face recognition. Firstly, the high-resolution (HR) and low-resolution (LR) training samples are respectively mapped into two different high-dimensional feature spaces by using kernel functions. Then CKFDA learns two mappings from the kernel images to a common subspace where discrimination property is maximized. Finally, similarity measure is used for classification. Experiments are conducted on publicly available databases to demonstrate the efficacy of our algorithm.