A research of eyes location algorithm based on expanded Haar features and fast Hough Transformation

Abstract

Eyes location is an essential part of human computer interaction. The precise of eyes location decides the feasibility of interaction. In this paper, we expand the haar features by introducing a new type of characteristic rectangle which can distinguish the eyes and brow more efficiently. The edge of iris is detected and recovered during preprocessing procedure. By using the distance information, the radius of iris can be predicted and the calculation of Hough Transformation can be optimized. By comparing the two improved algorithms, the one which constrained the range of radius can balance the detection accuracy and calculation efficiency better. This improved algorithm has a good performance on eyes location by enhancing the calculation efficiency one order of magnitude.