Eye state detection of driver based on monocular vision

Luhe Wan,Bingrong Hong,Zesu Cai

2008-01-01

Abstract:Driver's fatigue factors have been one of the most important causes of traffic accident, so driver fatigue monitor considered as the key technology of Safety Driving Assist technology has been paid more attention to it. Driver's eye state detection is a key problem of driver fatigue monitor. In this paper, we first use skin color Gauss model to locate the faces accurately and use Sobel to detect the feature of the faces, then we adopt elliptic moment function to locate the eyes roughly, further we locate it accurately by removing eyebrow factors according to the geometric position and then extract long axis and short axis features from the eyes. At last, we propose a method combined by statistical methods and the geometric characteristics to detect the three states of the eyes.