基于嘴部内轮廓特征的疲劳检测

针对传统利用嘴部开合度检测疲劳时容易发生嘴部定位误差且易受唇厚度影响,提出一种基于嘴部内轮廓特征的疲劳检测方法。首先改进了嘴部定位方法,将YCb Cr模型与Lab模型结合,去除类唇色干扰信息,利用a分量对唇色的聚类性定位嘴部,提高了定位的准确性;然后两次提取开合度优化哈欠特征的识别方法,提取嘴部开合度,过滤掉开合度较小的图像,对开合度较大的疑似哈欠图像做Gabor变换,提取嘴部内轮廓的开合度,修正唇厚度引起的误差,判断是否为打哈欠特征。最后根据具有局部连续性的哈欠特征出现的频率做出疲劳判决,降低了哈欠的误判率,提高了疲劳判别的可靠性。实验结果表明,该方法可以克服嘴部定位不准确及唇厚度的影响,有效地实现疲劳检测。

Fatigue Detection Based on The Inner Profile Characteristics of The Mouth

For the problem that the mouth opening degree judgment for fatigue driving is influenced easily by the effect of the lip thickness,a fatigue driving detection method based on the inner profile characteristics of the mouth which can be used to judge whether the driver is fatigued or not is proposed in this paper.Firstly,the YCbCr color model combined the Lab model is used to remove the non-skin regions in order to avoid the interference of the class color information.Used the clustering property for lip color of the component a in Lab color space to improve the accuracy of the positioning for the mouth;Secondly,used the mouth opening extraction for twice to improve the recognition method of yawn features,extracted the mouth opening degree and filtered out the smaller degree of non-yawn images,the Gabor transform is used to extract the internal contour feature of the mouth when the opening degree is larger and that corrected the error caused by the thickness of the lip,which can determine whether the driver is yawn or not;Finally, used the continuity characteristics of the yawn to judge the fatigue driving,which educed the false positive rate of yawn and improved the reliability of the fatigue driving determination.The results show that the proposed algorithm can overcome the non-accurate of the mouth position and the influence of lip thickness,achieved the driver fatigue detection effectively and has good accuracy.