Eye detection in face images for a driver vigilance system

Abstract

For a driver vigilance system one of the most important problems to be solved is eye detection. In this paper, we propose a new algorithm for eye detection that uses both iris geometrical information for determining in the whole image, the region candidate to contain an eye, and the symmetry for selecting the couple of eyes. The novelty of this work is that the algorithm works on complex images without constraints on the background, skin color segmentation and so on. After the first step, a validation algorithm based on a neural classifier has been applied to recognize if the couple of regions correspond effectively to eyes or false positives have been detected. Different experiments have been carried out on images of subjects with different eye colours, some of them wearing glasses. Tests showed robustness with respect to situations such as eyes partially occluded. In particular when applied to images where people have the eyes closed, the proposed algorithm correctly reveals the absence of eyes.