摘要
传统方法通过预警得出的驾驶状态与实际状态不符,为了降低道路交通安全隐患,研究一种连续驾驶条件下基于保局投影的疲劳驾驶预警方法。设定图像采集流程,针对获取的图像,通过自适应中值滤波法确定滤波器运行规则,去除图像噪声,并采用自适应阈值光照均衡方法转变图像色彩空间,找出参考白点调整光照,提高图像质量;选择相应投影矩阵,利用准则函数将高维数据映射在低维特征空间中,获取最小化保局投影矢量,提取出最显著的疲劳特征信息;分别设置眼部、嘴部、头部的疲劳特征指标,根据这些指标判断驾驶员是否存在疲劳状态,若存在及时给出预警。仿真结果证明,上述方法在保证图像质量的前提下预警结果与驾驶员实际状态相符。
In order to improve road traffic safety and reduce the gap between the driving state obtained by traditional early warning methods and the actual situation, this paper studies a fatigue driving early warning method based on local projection under continuous driving conditions. The image acquisition process was set to acquire images. The adaptive median filter method was used to determine the filter operation rules to remove image noise. The adaptive threshold illumination equalization method was introduced to transform the image color space and find the reference white points to adjust the illumination for further improving the image quality. The corresponding projection matrix was selected. Based on the criterion function, the high-dimensional data were mapped into the low-dimensional feature space to obtain the minimum locality preserving projection vector, extracting the most significant fatigue feature information. The fatigue characteristic indexes of eyes, mouth and head were set, respectively. Then the driver’s fatigue state was judged via these indexes. Once the driver is in a fatigue state, the system will give early warning information. The simulation results show that the early warning results of this method are consistent with the actual state of the driver.
作者
李巧茹
石宇
LI Qiao-ru;SHI Yu(Hebei University of Technology,School of Civil Engineering and Transportation,Tianjin 300400,China)
出处
《计算机仿真》
北大核心
2022年第6期175-178,254,共5页
Computer Simulation
关键词
连续驾驶
保局投影
疲劳预警
中值滤波
准则函数
Continuous driving
Locality preserving projection
Fatigue warning
Median filter
Criterion function
