Research on Fatigue Driving System Based on OpenCV

This paper proposes a human eye recognition algorithm based on Haar-like feature and adaboost classifier. Aim to solve the problem that traditional human eye detection technology have in low recognition rate and long training time. The first thing to do was to position the face. After the face was positioned successfully, we used the upper half part of the face image to further locate the human eye. Then extracted the haar features and trained human eye classifier by using adaboost algorithm. This experiment shows that method used in the paper has higher recognition rate than the traditional adaboost algorithm. And it can effectively reduces the interference effect of light in the recognition process.