Eye-blink Detection Using Gradient Orientations

State-of-the-art algorithms offer a real-time human face and eye detection which can be used to detect eye-blink. In this paper we present a feature descriptor. Gradient orientations and magnitudes are used to build the feature descriptor. We use the difference between samples of open and closed eyes. Gradient descriptors are further used to train the SVM. SVM can predict the open or closed eye state from the given input data. Due to the knowledge of the state of the eye (open or closed), eye-blink frequency and duration can be computed. These parameters are used to establish the level of sleepiness what can be used to prevent the driver from incoming microsleep.