Eye state detection for driver inattention based on Lucas Kanade optical flow algorithm

Traffic accidents due to drivers inattention has become one of the major factor in road accidents and highway crashers. The problem has led many researchers to develop drivers monitoring systems which can send warning signals to the drivers. One of the active research area related to this issue involve in determining the drivers mental state based on drivers' facial expressions which is non-intrusive as compared to analysis of brain wave, heart rate, excessive body heat. Apart from that, research on facial expression is expanding to the dynamic analysis whereby the temporal data is taken into consideration to determine the state of mind. Lucas Kanade optical flow is one of the methods that is being used to detect motions of objects. The method is later being proven to be capable of detecting facial features motion. However, there are no specific research being done on sleepiness and eye region motion analysis using Lucas Kanade optical flow method. This paper presents an optical flow using the Lucas Kanade algorithm by measuring optical flow on eye region from video sequences. In this stage, the drivers face and eye region are detected using facial feature detector. The optical flow of the eyes region is then computed. Results show that the proposed method can be applied to extract significant temporal data from the eye region.