基于嘴部内轮廓特征的疲劳检测

发布时间：2018-01-25 05:39

本文关键词： 疲劳检测唇部检测特征提取　出处：《科学技术与工程》2016年26期 　论文类型：期刊论文

【摘要】：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。
[Abstract]:In view of the traditional fatigue detection using mouth opening and closing degree, it is easy to occur mouth location error and easy to be affected by lip thickness. A fatigue detection method based on mouth inner contour feature is proposed. Firstly, the mouth location method is improved. The YCb Cr model is combined with the Lab model to remove the lip color interference information and to use a component to cluster the lip color to locate the mouth, which improves the accuracy of the location. Then the open and close degree is extracted twice to optimize the yawning feature recognition method, the mouth opening and closing degree is extracted, the smaller open and closed image is filtered out, and the suspected yawning image with high opening and closing degree is transformed by Gabor. The opening and closing degree of mouth contour is extracted, the error caused by lip thickness is corrected, and the yawning feature is judged. Finally, fatigue judgment is made according to the frequency of yawning feature with local continuity. The error rate of yawning is reduced and the reliability of fatigue discrimination is improved. The experimental results show that the method can overcome the influence of the inaccurate location of the mouth and the thickness of the lip and realize the fatigue detection effectively.
【作者单位】：河北工业大学电子信息工程学院;
【基金】：河北省高等学校自然科学研究重点基金(ZD20131043) 天津市自然科学基金重点项目(14JCZDJC32600)资助
【分类号】：U463.6;TP391.41
【正文快照】： 研究重点基金(ZD20131043)和天津市自然科学基金重点项目(14JCZDJC32600)资助现代社会中,汽车被视为科技进步的一种标志,方便了人们的出行与工作,但是,频发的交通事故严重威胁着驾驶员与乘客的生命安全。2015年,世界卫生组织道路安全现状报告显示,2007~2013年期间,每年有125万

【相似文献】