摘要
针对驾驶员疲劳检测算法中数据量大、高速传输、复杂运算的实际需要,以DSP器件TMS320DM642为核心处理器,开发了嵌入式的驾驶员疲劳驾驶状况实时监测系统;为解决因疲劳/瞌睡驾驶而造成的交通事故,针对国内各种疲劳检测方法大都采用单一的疲劳特征进行疲劳识别的现状,运用模糊神经网络方法,将多个疲劳特征参数:眼睛闭合时间占某特定时间的百分率(PERCLOS)、眼皮的平均闭合速度(AECS)、点头频率(NodFreq)、哈欠频率(YawnFreq)结合起来对驾驶员疲劳状况进行识别,准确率达88.7%.试验结果表明,算法对疲劳检测问题有较好的效果,系统的开发对降低因驾驶疲劳引发交通事故发生率的研究具有重要意义.
Aimed at the needs of large data, high transmission speed and complex operation, an embedded real-time monitoring system of fatigue driving is developed based on DSP TMS320DM642. In order to reduce the crash accidents caused by fatigue and drowsiness, various fatigue detecting methods are investigated. Fuzzy neural network is used for detecting driver fatigue status, combined with multiple fatigue characteristic cues such as: PERCLOS, AECS, NodFreq and YawnFreq, and the accuracy rate is 88.7%. The results show that the algorithm has preferable effect on fatigue detecting. The developed system has great significance in reducing incident rate of accidents for driver fatigue.
出处
《江苏大学学报(自然科学版)》
EI
CAS
北大核心
2008年第2期123-126,共4页
Journal of Jiangsu University:Natural Science Edition
基金
江苏省汽车工程重点实验室开放基金资助项目(QC200402)
江苏省图像处理与图像通信重点实验室资助项目(ZK2004002)
江西省科技厅2006年科技项目