A robust system for eye state recognition using the Hamming distances of eye image intensities

Eye state recognition is still a challenging work in the field of computer vision. Many researchers have described their methods which can work well with frontal face views, but not with variations of head poses. Some have explained that their methods deal effectively with head pose problems, yet the whole system is complex to implement and consumes a lot of processing time. In this paper, a novel and robust method for eye state recognition is presented in order to overcome the head pose and time complexity matters. Moreover, the system which can be simply divided into three main steps is easy to implement. The idea is to use machine learning to localize nose tip, eye regions, and then classify eye states according to the summation of many Hamming distances. A Hamming distance is computed between two observed eye image intensity profiles, which stay next to each other along the vertical axis. Since eye opening always scores higher than eye closing, this enables eye states to be distinguished well. Two databases are used for evaluating the proposed method, and the experimental results show that more than 98% in average of accuracy can be achieved, with time complexity of 12 milliseconds, for a single 384 by 286 resolution image.