Drowsiness Detection Using Eye-Blink Pattern and Mean Eye Landmarks’ Distance

Abstract

Road accidents have become a common phenomenon in this modern era. Reasons for road accidents are many. Driver drowsiness can be considered one of the major reasons. It creates a distraction which may lead to a road accident. For reducing the frequency of road accidents, effective steps should be taken to reduce driver drowsiness. Here we have brought a noble automatic method to detect the drowsy driver from real-time video monitoring. This proposed approach is a combination of image processing techniques and machine learning algorithms. The algorithm mainly analyses the eye blink pattern and mean eye landmarks’ distance of the drivers. The frequency of eye blink becomes low if drowsiness occurs. The mean eye landmarks’ distance is used to differentiate between the open eye and closed eye. In order to spot the sleepiness of the driver, firstly the face and then the eye of the driver are correctly detected. From the detected eye the facial landmarks’ position around the eyes is determined and from the eye landmarks’ position, the mean eye landmarks’ distance and thus the eye state is determined. If the eye is closed, then the duration of time for the closed state is considered to determine the drowsiness condition. If the duration is high, for giving warning to the driver an alarming system is attached.