Detection of Micro Movements of Facial Muscles in Facial Videos

During the last decade there has been many efforts by researchers to investigate the micro motion magnification from different types of videos. This was mainly due to its impact in many computer vision applications. In this paper we present how to detect these micro movements in natural face videos using basic band pass filtering. We then propose to utilize Complex Wavelets to magnify detected micro movements and make then visible. This proposed technique when applied on natural face videos can reveal different muscle movements which would indicate many different symptoms and expressions. Our proposed technique can be utilized in Forensic science, automatic deception detection, medical diagnose and psychotherapy. Several experiments on a large Face video database are reported along with results analysis and comparisons with similar literature techniques.