Suspicious Face Detection based on Eye and other facial features movement monitoring

Visual surveillance and security applications were never more important than now more so due to the overwhelming ever-growing threat of terrorism. Till date the large scale video surveillance systems mostly work as a passive system in which the videos are simply stored without being monitored. Such system will be useful for post event investigation. In order to make a system that is capable of real-time monitoring, we need to develop algorithms which can analyze and understand the scene that is being monitored. Generally, humans express their intention explicitly through facial expressions, speech, eye movement, and hand gesture. According to cognitive visiomotor theory, the human eye movements are rich source of information about the human intention and behavior. If we monitor the eye movement of a person, we will be able to describe him as an abnormal suspicious person or a normal person. We track his/her Eyes and based upon the eye movement in successive frames of the input videos using the Non-linear Entropy of eyes. Results of our experiments show that Non-linear Entropy of Eyes of an abnormal person is much higher than the eye's entropy of any normal person.