基于EOG的安全辅助驾驶系统算法设计与实现  

Design and implementation algorithm of safe driver assistant system based on EOG

作　　者：吕钊[1,2] 吴小培[1,2] 张超[1,2] 卫兵[1,2] 

机构地区：[1]安徽大学信息保障技术协同创新中心,安徽合肥230601 [2]安徽大学计算机科学与技术学院,安徽合肥230601

出　　处：《通信学报》2016年第7期87-95,共9页Journal on Communications

基　　金：国家自然科学基金资助项目(No.61401002;No.61271352);安徽省自然科学基金资助项目(No.1408085QF125);安徽高校省级自然科学研究重点基金资助项目(No.KJ2014A011)~~

摘　　要：为保证驾驶安全,提高车辆控制系统的智能化水平,实现"手不离盘"操作,设计并实现了一种基于眼电图(EOG)的安全辅助驾驶系统。该系统利用安装在驾驶员眼睛周围的生物电极采集其在观测抬头显示器(HUD,head up display)上提示符时所产生的扫视信号,生成多种车载设备控制命令;对原始多导联EOG信号进行端点检测后,使用了独立分量分析(ICA,independent component analysis)方法进行空域滤波后提取眼动信号特征参数,并结合支持向量机实现了上、左与右扫视动作的识别。实验室环境下对所提算法进行了测试,15位受试者在疲劳与非疲劳状态下的在线平均正确率达到了98.43%与96.0%。实验结果表明,基于ICA多类扫视信号识别算法的安全辅助驾驶系统在眼动信号分析中呈现出了良好的分类性能。In order to ensure driving safety, improve the intelligent level of the vehicle control system and realize ＂keeping hands on the wheel＂, a safe driver assistant system（SDAS） based on EOG was proposed. The proposed system utilized saccade signals which come from bio-electrodes installed around driver＇s eyes, to generate some control commands when the driver observes different signs located on the head up display（HUD）. Furthermore, independent component analysis（ICA） algorithm was used to extract spatial feature parameters of activity-detected EOG signals, and combined with support vector machine（SVM） method to recognize the type of saccade signals, such as up-rolling, left-rolling and right-rolling. Experiments have been carried out in lab environment, and the average correct ratio on 15 subjects is 98.43% and 96.0% corresponding to fatigue condition and non-fatigue condition respectively. Experiential results reveal that the SDAS based on the multi-class saccade signals recognition algorithm presents an excellent classification performance.

关 键 词：眼电图 扫视信号 独立分量分析 空域滤波 支持向量机 

分 类 号：R318[医药卫生—生物医学工程]