Human Fatigue Detection Using PPG Of Facial Skin

Human fatigue detection,on the one hand,can reduce security risks caused by fatigue driving,on the other hand,can discover sub-health state when people feel tired.Thus the non-contact,convenient detection of human fatigue has far-reaching significance and important research value.The related theory and technology of heart rate and heart rate variability extracted from video,detection of human fatigue through video are summarized and studied in this thesis.The principle and most extraction methods of photoplethysmography(PPG)signal are researched.The blind source separation methods to extract heart rate from video are deeply explored.Comparative analysis of results of extracted heart rate is carried out.Human fatigue detection approaches based on facial skin PPG is then proposed.The major work and innovations in this thesis are as follows.(1)The principle and simulation experiment of ICA methods to extract PPG signal of facial skin.The in-depth analysis and comparison of joint approximate diagonalization(JADE),diagonal cumulant and Robust ICA algorithm is carried on.The simulation algorithm of these three algorithms is designed.The separation of the mixed signal constituted by blind source using these three motheds can be best done,and performance in separation has different advantages.It does preparation for further research of this paper.(2)The extraction of heart rate based on PPG signal of skin color.Research on using diagonal cumulant,Robust ICA algorithm to separate PPG signal from video.Comparative experiments show that the absolute value of statistical characteristic parameters of heart rate value using diagonal cumulant and the real heart rate are lower than JADE,it suggests that diagonal cumulant algorithm has higher estimated accuracy.The running time of Robust ICA algorithm to extract heart rate is controlled within 0.2s,it suggests that using Robust ICA algorithm to extract heart rate from color video has a lower time complexity.The model of extracting PPG signal from color video is improved in this paper,skin color pixels detection is joined in face detection,the experimental results show that the accuracy of heart rate results can be improved by joining skin pixels detection.At the same time,in order to solve the disorder of extraction results using ICA,the method based on correlation to screen PPG signal is put forward,the PPG signal of low noise ratio can be eliminated,data is simplified,the accuracy of results and the speed of computing is improved.(3)Human fatigue detection based on PPG signal of skin color.The extraction of heart rate variation is realized by filtering,interpolation,peak detection and power spectrum estimation and other signal processing technology based on PPG signal of skin color extracted throught diagonal cumulants algorithm.The situation that the analysis of heart rate variation can only be done in the laboratory using complex instruments is changed.Human fatigue detection experiment based on the variation trend of heart rate and heart rate variation is designed,experimental results show that the trend of heart rate,heart rate variability can detect human fatigue.