The real-time eye-controlled system based on the near-field video streams

Eye movement is a direct expression of human's visual intent. The paper presents a novel system which is focusing on utilizing eye movement information as control source and the video streams from a remote pan-tilt-zoom (PTZ) camera as feedback. By processing left and right eye images simultaneously, the system can recognize the direction of eye movement and adjust the motion of remote PTZ camera to synchronize with the eye movement. Meanwhile, the user who is wearing video glasses (head-mounted display) can see the images captured by the PTZ camera. Besides, a robust algorithm is also put forward to overcome the interferences caused by the instability of background brightness and ways of wearing video glasses.