REAL - TIME FACE RECOGNITION USING EIGENFACES

Access control systems are used to grant or deny the access to a person of a particular resource. There has been an enormous change in the trend of access control systems in recent times. Starting with the use of physical access control systems such as tok ens, passwords etc., for the identification of a person, the trend has swayed towards designing and deploying of access control systems which use biometric identification of individual persons, for the grant or denial of access to resources. Biometric identification methods use various sources like retina, fingerprint, DNA etc., Biometric sources can be classified into two, namely physiological and behavioral. The former includes face, fingerprint, hand, iris, DNA and the latter includes keystroke, sig nature and voice. The access control systems using these biometric sources fundamentally identify and recognize a particular personal trait of person and compare it with the information available to grant or deny access to such person who seeks to interact with such system. As amongst such biometric sources to develop reliable access control systems researches have shown overt interest in using the face of a person (face recognition). Such inclination of the researchers is due to the various strategic advan tages face recognition systems have like, its global application, wide and compatible collectability of data, cost effectiveness in implementation (for example existing surveillance cameras can be used to deploy such systems) when compared to other biomet ric methods and many more. The current project fundamentally aims at successfully designing and implementing a face recognition technology to develop an access control system. Out of the various available methods for developing a face recognition technol ogy such as Fishersface, Hidden Markov model, dynamic link matching, three - dimensional face recognition, Eigenfaces etc., this project adopts Eigenfaces method of face recognition to achieve such aim. The project fundamentally detects and identifies human faces that work as a biometric source. This project aims to provide solution for the development of face recognition system by assuming that, problems involved in developing such systems are intrinsically a two dimensional rather than a three dimensional. It basically identifies the face of a person in a face image and then identifies the specific characteristics of such face image, then compares such characteristics with an existing database containing specific characteristics of several faces of differen t individuals, to decide if the former matches with any of the existing faces in the database. Eigenfaces method is utilised to achieve the above result where, face image is projected onto a feature space that spans the significant variations among known f aces. The best variation among different images is calculated where during such calculation not exactly the facial features like eyes, nose, ears, etc., are classified. Instead it learns each face in an under constant observation as a whole.