Gaze estimation using a webcam for region of interest detection

In this paper, a real-time gaze estimation system using a webcam is proposed, in which variation of head pose is tracked. At first, variation of head position and pose are estimated by using facial features. Then, an iterative iris center detection method is proposed for tracking iris in eye image. Finally, gaze is estimated by using estimated head pose and position, and iris center position. The proposed gaze estimation system is applied to four different applications. Experimental results show that the proposed iterative iris center detection method has a higher accuracy than conventional ones. Also, the proposed gaze estimation system shows about 98% accuracy using 640 x 480 resolution webcam and 42-inch monitor that are 0.75 m apart.