Face Recognition at a Distance

Abstract

Face recognition at a distance is generally motivated by the desire to automatically recognize noncooperative subjects over a wide area. This remote biometric collection and identification problem has been addressed with high-resolution stationary cameras and active camera systems. Key challenges include optical system design, pan-tilt-zoom camera targeting and control, and face recognition with low-resolution images and no pose or illumination control. We discuss major applications, challenges and approaches in this field, and review research literature on this and closely related topics. We further describe a specific face recognition at a distance system that uses the active camera approach, algorithms for facial image modeling and alignment for low-resolution images, and a multi-frame super-resolution process for facial images.