A fast eye state computing algorithm

Nowadays, the driver drowsiness has become a main cause of many traffic accidents, while the state of eye is a significant factor which has a close relationship with sleep. This article presents an algorithm which can estimate drivers´ eye states, aiming at reducing the rates of traffic accidents by monitoring drivers´ drowsiness. Firstly, a driver´s face images are grabbed from a digital camera and the Haar-like algorithm is implemented to extract the face area from these images. Then, the algorithm that we proposed can find the exact location of eyes and this step uses the technique called Integral Projection and several morphological operations, which makes the computing very fast. Finally, the eye states can be estimated by comparing the parameters of eye´s contour. Also, a prototype system is conducted to test the performance of the algorithm. It turns out that this algorithm is very efficient and can reach a high accuracy in different light conditions.