Face recognition using a kernel fractional-step discriminant analysis algorithm

Feature extraction is among the most important problems in face recognition sys-tems. In this paper, we propose an enhanced kernel discriminant analysis (KDA)algorithm called kernel fractional-step discriminant analysis (KFDA) for nonlinearfeature extraction and dimensionality reduction. Not only can this new algorithm,like other kernel methods, deal with nonlinearity required for many face recognitiontasks, it can also outperform traditional KDA algorithms in resisting the adverse ef-fects due to outlier classes. Moreover, to further strengthen the overall performanceof KDA algorithms for face recognition, we propose two new kernel functions: cosinefractional-power polynomial kernel and non-normal Gaussian RBF kernel. We per-form extensive comparative studies based on the YaleB and FERET face databases.Experimental results show that our KFDA algorithm outperforms traditional ker-nel principal component analysis (KPCA) and KDA algorithms. Moreover, furtherimprovement can be obtained when the two new kernel functions are used.