摘要
疲劳驾驶导致汽车交通事故逐年增加,为了提升驾车的安全性,需对驾驶员疲劳状态实时监测并及时提醒.为了提高疲劳驾驶判断效率及准确率,本文运用Viola-Jones框架特征矩阵进行人脸预判断;预判断过程中为了减少Haar值计算量并提高人脸识别速度,采用Adaboost算法和级联分析,剔除非人脸的Haar特征值,实现快速人脸识别;根据色彩空间转化实现眼部分割处理,根据PERCLOS值评估驾驶员是否处于疲劳状态并提前予以警示;通过MATLAB仿真软件实现疲劳驾驶检测算法的仿真分析.在多个样本的测试过程中,该方法有效识别出人脸,并能够准确监测驾驶员的疲劳状态.
Fatigue driving has increased the number of automobile accidents year by year.To improve the safety of driving,the driver's fatigue status must be monitored in real time and promptly reminded.To improve the efficiency and accuracy of fatigue driving judgment,this paper uses the Viola-Jones frame feature matrix for face pre-judgment.To reduce the calculation amount of Haar value and improve the speed of face recognition during the pre-judgment process,Adaboost algorithm and cascade analysis are used to eliminate the Haar feature value of the face to achieve fast face recognition.Eye segmentation is implemented based on color space conversion,and the driver is evaluated for fatigue and warned in advance based on the PERCLOS value.The simulation analysis of fatigue driving detection algorithm is realized by MATLAB simulation software.During the test of multiple samples,the method can effectively recognize the human face and accurately monitor the driver's fatigue state.
作者
吴雪颖
吴才硕
黄文聪
覃舒琳
WU Xueying;WU Caishuo;HUANGWencong;QIN Shulin(School of Vocational and Technical Education,Guangxi University of Science and Technology,Liuzhou 545006,China)
出处
《广西科技大学学报》
2021年第1期49-54,共6页
Journal of Guangxi University of Science and Technology
基金
广西高校中青年教师科研基础能力提升项目(2019KY0374)
2019年自治区级大学生创新创业计划训练项目(201910594133)资助。