Recent trends and challenges in real time face recognition system in video

Abstract

Summary form only given. Video surveillance systems are used to monitor the activities in the surveillance area. The existing video surveillance systems are reactive where the analysis is done after the occurrence of an event. Thus, there is a need for a proactive solution in real time to avoid the damage caused due to the occurrence of abnormal events. Face detection and recognition in online video streams plays a challenging role in surveillance system due to variation in illumination, pose and occlusion. This keynote address presents the methodologies for efficient face recognition invariant to Pose, Illumination and Occlusion. Video based face recognition system improves decision making in identifying the unauthorised person by considering continuous frames and temporal information. The video based face detection and recognition involves four modules such as face detection, pose estimation, feature extraction and face recognition. The modules are proved to be efficient in terms of accuracy and detection time.