Unauthorized access detection system to the equipments in a room based on the persons identification by face recognition

Face recognition is a very active research topic due to the number of potential applications that use this technique and the number of challenges that still require efforts to solve them. However, new research problems arise when dealing with devices operating in real-time, and using a non-controlled image source such as surveillance cameras, due to the needs and context of end-use. A robust and efficient face recognition system is proposed in this paper to monitor unauthorized access to equipments in a room using a surveillance camera. The system is based on a person detection algorithm adapted to the context of use, a person tracking algorithm robust to occlusion, a face detection algorithm robust to orientation and scale, and a face recognition algorithm robust to illumination change. To enhance the efficiency of person identification, a novel strategy for estimating the confidence of face recognition is proposed in this paper. This is achieved by leveraging the general conditions of the image used for this purpose. This technique allows to increase the identification rate while minimizing the false alarm rate. An effort is also made to optimize the implementation of the developed algorithms to guarantee a real-time operation while keeping a very high efficiency. Also, the evaluations of the proposed algorithms on public and private data show an improvement of the accuracy on the whole context of use of the proposed system. Furthermore the evaluation of the system in general shows a high efficiency, 100% detection of unauthorized accesses and 0 false alarms.