Face Recognition using Fusion of Local Binary Pattern and Zernike Moments

Numerous approaches are developed for the face recognition and its related applications. These approaches are solely dependent upon the kind of feature vectors used for extracting the different characteristics and structural contents from the image. The ability of extracting features from the image is either local or global. In this paper, we present a fusion of local features i. e. Local Binary Pattern (LBP) with global features i. e. Zernike moments (ZM) to provide the more reliable and robust face recognition system. The effectiveness of proposed fused feature vector scheme are tested on FERET and ORL database and its performance is observed by using different classifiers such as Euclidean distance, Chi-square, Square chord, histogram intersection and Canberra distance measures. The experimental results clearly state that the proposed fusion of LBP with ZM with Chi-square and Canberra as distance measure provide the better results than individual LBP and ZM.