Approach towards real time face recognition in streaming video under partial occlusion

Abstract

Security system basically relies on biometrics. which gain vital importance for implementing a security systems. Biometrics such as iris, fingerprints, face, etc. Among this biometrics face is most prominent and trust worthy for implementing security system. It plays a vital role in identification and verification in various security based system. Face recognition is widely considered as one of the most promising biometric techniques. It allows high recognition rates without being too intrusive. Generally face recognition is used for two primary tasks: 1. verification (one-to-one matching. 2. Identification (one-to-many matching): Many approaches has been presented to solve the problem of face recognition in still images but a very less work is carried on face recognition in streaming video under uncontrolled condition. This paper illustrates various issues on face recognition in streaming video under, partial occlusion. Occlusion refers to any hindrance in the view of face. Occlusion can be natural and artificial. Natural occlusion refers to the facade of the face image which is differ from original face image caused due to natural calamity or disasters. whereas artificial occlusion refers to the artificial blockage of intentionally covering the face image's view by artificial accessories like sunglasses, makeup, hair wrapping, hands, scarf's, etc. Further on the basis of face recognition approach in presence of partial occlusion a real-time system for recognizing faces in a video stream provided by a surveillance camera was implemented, having real-time face detection.