W. Yu, L. Deng, Y.H. Wu

Pages: 15-26

Abstract
Fatigue driving refers to the imbalance of physiological and psychological functions after driving for a long time, which objectively leads to the decline of driving skills, which is very easy to cause traffic accidents. In order to improve the recognition rate of fatigue driving behavior, a fatigue driving behavior recognition method based on image and vehicle feature changes is proposed in this paper. Firstly, the artificial skin color change of tired drivers is determined according to the color space; Then, the color image is grayed, and the gray image is enhanced by median filter; Then, the nonlinear signal features of vehicle feature changes are determined by using probability density function and probability distribution function, the features of fatigue driving vehicles are extracted, and the mathematical model is established. Finally, with the help of the changes of skin color image and vehicle characteristics of fatigue drivers, the K-means fuzzy clustering method is used to calculate the feature membership, and the fatigue driving behavior recognition is realized according to the maximum membership. The results show that the recognition rate of fatigue driving behavior is 97.4%.
Keywords: image processing; changes in vehicle characteristics; fatigue driving; behavior identification; principal component analysis