Eye detection to assist drowsy drivers

This paper proposes a new approach to detect the eyes opening state for the purpose of alarming drowsy drivers on highroads. The input image frame is processed in successive steps using a combination of algorithms, namely wavelets transform, edge detection and YCbCr transform. The algorithm performs well, regardless the size of the image and is immune to distracting noises, high luminance, and background lighting. The proposed algorithm was able to attain an average eye detection accuracy of 80% when tested on different data bases and different conditions.