A Real-time Pattern Detection using Fast Feature Matching Algorithm

A real-time eye tracking using fast face detection algorithm is proposed in this paper. Most of the current eye tracking system has operational limitations such sensors, complicated backgrounds, and uneven lighting condition. It also suffers from slow response time which is not proper for real-time applications. We propose a fast feature matching algorithm to track eyes for real-time purpose. The base of the proposed algorithm is to use Ellipsoidal Hough Transformation to detect face region within images followed by the detection of eye area by computing and segmenting a limited set of Haar-like features within the detected face region. To speed up the algorithm execution time, the algorithm uses motion estimation to limit the search area, reduction of the number of edge pixels using non-maximum edge suppression, limitation of scale factor using Region-of-Interest finder, and hierarchical processing. Detected eye location is fed into the 3-dimensional display system to properly show 3-dimentional images in real-time.