Face and eye tracking for controlling computer functions

This paper presents a novel user interface with computer. We propose eye and face detection via Haar-like feature with Bilateral Total Variation based super-resolution to resize the low resolution eye image from web camera. The center of eye-pair detection is mapped to the mouse coordinates in computer screen. Left click and right click are controlled by blinking left eye and right eye, respectively. Ambient illuminations are fluorescence and natural daylight in day time mode and fluorescence as one light source at night time mode. The system performance are evaluated on measuring sensitivities of face and eye detection across (1) distances between web camera and the user, (2) roll and pitch angles of the head pose, and (3) eye-center coordinates for controlling computer functions.