Fusion of local and global features for face recognition

In recent years, there have been lots of research developments done in face recognition systems. Face recognition systems are widely used for access control, border control, surveillance and in law enforcement. Among other biometrics, it is the most natural and acceptable way of identifying an individual. Face recognition system does not require physical interaction with the user. Research is still being done intensively to produce systems that can cater for several challenges such as changes in pose, illumination, occlusion and low resolution images. Algorithms reported in literature use either global feature extraction or local feature extraction. In this work, a different technique is proposed that combines both global and local approaches for face recognition. The Principal Components Analysis (PCA) and Local Binary Patterns (LBP) have been employed. Face recognition yields a recognition rate of 90 % with PCA and 92 % with LBP. However, results show an improvement in recognition rate to 95 % when both approaches are fused.