A Mental Fatigue Detection System Based On Multi-mode Physiological Signals

Physical fatigue is one subjective and uncomfortable feeling because of prolonged physical activities or overwork.Fatigue can lead to a decline in physiological and mental function,which can affect our attention,perception,thought,judgement,volition,decision and exercise.In our daily life,there are a lot of negative influence due to fatigue,such as fatigue during driving which can lead to traffic accidents easily,and fatigue in learning process which may cause can lead to a low study efficiency,and so on.Therefore,to develop a method which can detect fatigue efficiently and accurately is particularly important.In this paper,a mental fatigue detection method based on ECG,EOG and eye movement signals is proposed,and its hardware circuit and corresponding algorithm are designed,also,we finished its performance test.Original ECG,EOG or eye movement signals can not represent fatigue state perfectly.However,in this paper,we calculate heart rate variability(HRV)through measuring ECG,gain blinking frequency through EOG signals obtained from the single conducting electrode at the frontal lobe,and detect perclos through a camera and open CV image processing library.Any one of the three physiological index: HRV,blinking frequency and perclos can represent test subjects' mental fatigue well.In this paper,three fatigue state parameters are calculated respectively based on these three characters.Finally,the final fatigue state index is acquired though a simple decision layer multi-mode fusion based on the rough set theory.In this design,firstly,the collection principle of these three signals is elaborated,and the corresponding acquisition circuit is designed respectively.A specialized week signal collector is designed for the ECG and EOG signals.Strict noise and calculation and control are carried out in the front-end of the collector's analog sampling,also its power supply and main control unit are also finished.To ensure the accuracy of its result,a de-nosing algorithm is designed,which is to eliminate the baselines and power frequency.Then,these processed signals will be transmitted to the control terminal based on an arm processor via bluetooth in real time.Finally,HRV can be calculated by the differential sequence spectrum from R waves' time interval,the blinking frequency is obtained by trigger detection and calculation.On the other hand,gain the facial images through a high definition camera controlled by ARM terminal,and then start a noise reduction processing and a brightness enhancement processing,and identify faces and eyes by AdaBoost algorithm.Finally,a image binarization processing can provide us a monochromatic bitmap,and its horizontal and vertical integral result is the closure degree of the eyelid.Compared with the traditional fatigue detection method based on a single feature,the system of this paper uses these several physiological parameters as a criterion of fatigue,and has a strict calculation and control to the acquisition circuit' noise and interference,which can lead to an improvement of its real-time performance and accuracy.To prove this method' feasibility,10 testee are chosen,who are induced to enter the fatigue state by the neutral video stimulus.During the tests,these testee' physiological parameters are acquired by this fatigue detection system,which is compared with the parameters acquired at the ten minutes earlier before the test or ten minutes later after the test.The conclusion is: the method is feasible,the compatible hardware and algorithm system is worked availably,and the whole system can determine fatigue effectively.