基于人眼特征的疲劳驾驶检测技术研究

Research on the fatigue driving detection technology based on eye features

针对目前疲劳驾驶检测技术无法很好地平衡算法准确性和实时性的问题,本文在人脸特征点定位方面采用了实时性较好的基于回归的局部二值特征法(LBF算法)。同时为了提高检测精度,改进了该算法的初始化策略。并且在建立LBF随机森林时使用归一化的像素特征取代原始特征,从而提升分类效果。通过眼部宽高比检测人眼闭合程度,并提出将人眼视线方向应用于疲劳驾驶检测算法中,判断注意力是否分散,以便在驾驶员陷入深度疲劳之前对其预警。利用上述得到的多特征综合检测疲劳程度,实验结果表明该技术提高了算法的准确性。

In view of a fact that the current fatigue driving detection technology can’t balance the accuracy and real-time performance of the algorithm very well,in this paper,the real-time regression-based local binary feature method(LBF algorithm)is adopted for face feature point localization.At the same time,in order to improve the detection accuracy,the initialization strategy of the algorithm is improved.And when establishing LBF random forest,the original features are replaced by normalized pixel features to improve the classification effect.The degree of eyes closure is detected by the eye aspect ratio,and the direction of human eyes is applied to the fatigue driving detection algorithm to determine whether the attention is dispersed,so as to warn before the driver falls into deep fatigue.Using the multiple features obtained above to comprehensively detect the degree of fatigue,the experimental results are derived,showing that the technique improves the accuracy of the algorithm.