Gabor Filter Using on Graph Regularized Non-negative Matrix Factorization for Face Recognition

Abstract

Gabor Filter is a famous linear filter which is usually used for edge extraction.In this paper, we introduced Gabor Filter into Graph regularized Non-negative Matrix Factorization for face recognition.Thanks to the Gabor Filter, important information of the original picture can be highlighted so the recognition rate can be improved.We have performed a series of experiments on NMF algorithm with different Gabor Filter to exhibit the improvement on recognition rate.Next, we make some discussion on the accuracy of the neighbor graph construction method under different Gabor Filters.We also focus on the recognition rate of the Gabor Filter based GNMF algorithm with and without knowledge.Experiments demonstrate the effectiveness of our method.