Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

The Fatigue State Recognition of the Driver Based on Eye Detection

Author LiuJiangWei
Tutor HongBingZuo
School Harbin Institute of Technology
Course Computer Science and Technology
Keywords Detection of fatigue driving AdaBoost algorithm Kanade-Lucas Optical algorithm Ellipse fitting
CLC TP391.41
Type Master's thesis
Year 2008
Downloads 185
Quotes 1
Download Dissertation

As the vehicle population grows rapidly in China, the traffic safety problem is becoming increasingly important. And fatigue driving is one of the main factors that cause traffic accidents. So the fast and effective detection of the driver’s fatigue state can be a key solution to this problem. In this paper, we judge the fatigue state by detecting the state of the driver’s eyes using the method of computer vision.Firstly, we using the AdaBoost algorithm to detect the region of the driver’s face, and then detect the region of eyes in the face we find using the same method. With a relative small cost, we can gain a better detecting performance than detecting only once.Although the Adaboost algorithm is fast and can achieve a low false positive rate, it can also cause a relative high false negative rate on the contrary, and reduce the reliability of the system. So we use Kanade-Lucas optical algorithm to track the eyes at the same time. This method can reduce the danger caused by the false negative detection of the Adaboost algorithm. We proved that the K-L optical algorithm can be a effective complement of the Adaboost algorithm through experiment, and can increase the robustness of the system that make it useful in reality.After we find the region of driver’s eyes, we use the threshold segmentation and the Freeman chain code method to extract the contours of the eyes. Then we use the ellipse fitting algorithm to fit the eye contours. Finally, we judge the state of the eye by the threshold which derived from statistic data obtained by experiment. When the system detects the driver was tired it can trigger the alarming device to awake the driver.

Related Dissertations
More Dissertations