Face identification using local and global features

Extract facial feature is considered as the most important step in face recognition systems. Several researches have been made to improve facial feature extraction techniques. This paper proposes hybrid global and local feature based on Gabor filter bank and double coding local binary pattern (LBP). The proposed features are concatenated and reduced using generalized discriminant analysis (GDA). The matching step is then performed using Euclidian distance. Performances evaluation of face identification system is carried out using ORL and FERET databases, and the experimental results shown the high recognition rate of the proposed system.