摘要
为减少交通事故,采用基于数据融合的疲劳检测技术以提高疲劳检测精度.通过驾驶行为与车辆跟踪技术研究现状分析,选择眼睑遮住瞳孔的面积超过80%的P80和眨眼次数指标作为眼部特征参数、车辆越线指标作为驾驶行为特征参数.将两个特征参数分为3类,分别为:清醒状态、轻微疲劳状态、疲劳状态;最后通过支持向量机算法建立基于数据融合的疲劳检测模型.实验结果分别为灵敏度为86.45%,检测准确率为85.79%,特异度为84.63%,较单一数据源的疲劳检测方式精准,建立的融合模型提高了疲劳检测的准确性.
To reduce traffic accidents,we adopted fatigue detection technology based on data fusion to improvethe accuracy of fatigue detection. By analyzing the driving behavior and vehicle tracking technology,P80(theeyelids cover the pupillary area of more than 80%)and blink frequency were selected as the eye characteristicparameters,and the vehicle cross line was selected as the driving behavior characteristic parameters. The twocharacteristic parameters were divided into three categories,the waking state,mild fatigue and fatigue;finally,the fatigue detection model based on data fusion was established by supporting vector machine. Experimentalresults show that the sensitivity is 86.45%,the detection accuracy is 85.79%,and the specificity is 84.63%,which is more accurate compared with the fatigue detection method based on single data source. It is concludedthat the established fusion model can improve the accuracy of fatigue detection.
出处
《武汉工程大学学报》
CAS
2016年第5期505-510,共6页
Journal of Wuhan Institute of Technology
关键词
驾驶行为
疲劳识别
车道偏离
P80
支持向量机
数据融合
driving behavior
fatigue detection
lane departure
P80
supporting vector machine
data fusion