基于眼电与头部姿态信号的检测与疲劳状态分析

Detection and Fatigue State Analysis Based on EOG and Head Attitude Signals

当人处于疲劳状态时,其眼电信号特征及头部姿态信号特征均会发生明显的变化。针对这两类信号进行分析研究,提出一种可穿戴式眼电与头部姿态信号的疲劳检测装置。利用三个Ag/Agcl电极单导联方式采集人眼电信号、MEMS传感器采集人头部运动时的加速度和角速度信号。根据眼电信号及加速度和角速度在时域中的特点,利用相关系数分析左右电极所采集的眨眼信号特征,并根据加速度与角速度在时域中的特点分析四种头部姿态特征。最后利用BP神经元网络对眨眼信号及头部姿态信号进行特征识别,提高了检测系统鲁棒性。实验结果表明,利用水平眼电信号与加速度信号能准确分析测试人员的眨眼与低头、仰头行为,并能正确检测人的疲劳状态变化,但侧头行为的疲劳状态检测有待进一步优化提升。

When a person is in a state of fatigue,the characteristics of the EOG signal and the characteristics of the head posture signal would change significantly.In this paper,two types of signals were analyzed and researched;and a fatigue detecting device for wearable EOG and head attitude signals was proposed.The three-in-one Ag/Agcl electrode single-lead method was used to collect the human eye electrical signal;and the MEMS sensor collected the acceleration and angular velocity signals of the human head during motion.According to the characteristics of EOG signal and acceleration and angular velocity in the time domain,the correlation coefficient was used to analyze the characteristics of the blink signal collected by the left and right electrodes;and the four head pose features were analyzed according to the characteristics of acceleration and angular velocity in the time domain.Finally,the BP neural network was used to identify the blink signal and the head attitude signal,which improved the robustness of the detection system.The experimental results showed that the horizontal EOG signal and the acceleration signal could accurately analyze the tester's blinking and bowing and head-up behavior,and could correctly detect the fatigue state change of the person,but the fatigue state detection of the side-head behavior needed further optimization and improvement.