Real-time eye detection in multimedia environment

Eye detection is more and more important in many applications of Multimedia Environment, such as face detection, video surveillance, emotional analysis. Many methods relay on some expensive hardware, and they are also complicated. So in this paper, we propose a simple method only with a webcam to realize real-time eye detection. First, we seize human face from the webcam. And our program will search eyes automatically. Then, we convert eyes’ image to grayscale image in order to post-processing. We use Gaussian smoothing to reduce noise and Histogram equalization to enhance contrast. Finally, the image is binarized so the eye features are more prominent. And we can use Hough transform to draw a circle that fits the eyeball, so the center of the circle is the pupil. The experiment verifies the effectiveness of our method.