Distributed Face Recognition: A Multi-Agent Approach

In this paper we present an application of agent technology to the problem of face recognition. With a new composite model consisting of multiple layers, the system can achieve high performance in terms of robustness and recognition in complex visual environmental conditions. The robustness of the complex face recognition system is enhanced due to integration with agent based paradigm, with more than 95% accuracy achieved under illumination, pose and expression variations of faces in images with multiple faces, and background objects. The results of preliminary findings are promising, suggesting further investigations in intelligent agent methodology for facial biometrics using fusion of face, gait, gesture, and voice biometric traits to person identity recognition problem in distributed scenarios such as video surveillance, health informatics, and crime investigation applications.