Occlusion invariant face recognition using selective LNMF basis images

In this paper, we propose a novel occlusion invariant face recognition algorithm based on Selective Local Nonnegative Matrix Factorization (S-LNMF) technique. The proposed algorithm is composed of two phases; the occlusion detection phase and the selective LNMF-based recognition phase. We use local approach to effectively detect partial occlusion in the input face image. A face image is first divided into a finite number of disjointed local patches, and then each patch is represented by PCA (Principal Component Analysis), obtained by corresponding occlusion-free patches of training images. And 1-NN threshold classifier was used for occlusion detection for each patch in the corresponding PCA space. In the recognition phase, by employing the LNMF-based face representation, we exclusively use the LNMF bases of occlusion-free image patches for face recognition. Euclidean nearest neighbor rule is applied for the matching. Experimental results demonstrate that the proposed local patch-based occlusion detection technique and S-LNMF-based recognition algorithm works well and the performance is superior to other conventional approaches.