Pose-Invariant Face Recognition Using Subspace Techniques

Biometrics refers to the growing technology of automated methods in recognizing individuals based on their physiological or behavioral characteristics (e.g. fingerprints, irises, faces, etc.) to either verify a claimed identity (biometric verification) or establish the identity of an individual (biometric identification). Concerns about security, fraud and illegal access have prompted governments, airlines,banks and others to become more interested in the technology. Face recognition, in particular, hasgenerated much more interest than other biometric systems because of the multitude of applications it enables. Unfortunately, face recognition is susceptible to variations in pose, light intensity, expression, etc. In addition to a review of biometric techniques and current approaches proposed to solve the face recognition problem, we introduce a robust face recognition system, using subspace techniques including Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), to both estimate the pose and identify the individual. copy; 2005 John Wiley & Sons, Ltd.