Research On Human Vital Signs Perception Based On IPPG

Heart rate,respiratory rate,and oxygen saturation(Sp O2),as important vital signs of human body,are the three major indicators for the detection and diagnosis of cardiovascular and respiratory diseases,and have important reference value in personal daily health monitoring.The traditional contact-based vital sign sensing method requires direct contact with the skin of the tested subject,which can cause discomfort when used for a long time.Moreover,when tested patients with infectious diseases,it can bring a certain infection risk to the testing personnel.In contrast,the non-contact vital sign detection method based on imaging photoplethysmography(IPPG)technology can extract physiological signals from human facial images,and has the characteristics of non-contact,non-invasive,repeatable operation,low cost and high accuracy.It is of great significance for the popularization of personal daily home testing,large-scale group clinical testing and remote testing.However,at present,the IPPG source signals obtained by collected face videos with the common camera have noise interference,and contain weak physiological signals,so it is difficult to directly detect vital signs.Based on the above background,this paper studies the IPPG-based human vital signs perception system.The main contents are as follows:(1)In order to solve the problem of weak physiological signals extracted from facial videos,the Eulerian Video Magnification is used to enhance the color of human face videos,which increases the intensity of the physiological signal.In order to explore the influence of the region of interest(ROI)on the IPPG source signals,the signal quality obtained by a variety of different ROIs is compared and analyzed,and finally the nose and its nearby cheek region is determined as the ROI for extracting IPPG source signals.(2)Aiming at the problem that the IPPG source signals are disturbed by noise,a signal processing algorithm based on the combination of "wavelet transform-principal component analysis-fast independent component analysis" is proposed.Firstly,by comparing the decomposition characteristics of several algorithms,the decomposition method of wavelet transform is determined to reduce noise in the first step.Then,according to the characteristics of human physiological signals,the principal component analysis is used to reduce the dimension,after that the fast independent component analysis is used to extract the signal features.(3)In order to ensure the accuracy of the perception of each vital sign,the absorption characteristics of various light-absorbing substances in the skin are analyzed.Then,combined with the light absorption characteristics of oxyhemoglobin and deoxyhemoglobin in the red,green and blue color channels,the signal after noise processing of a specific color channel is selected for data analysis,and the non-contact measurement models of heart rate,respiratory rate and Sp O2 parameters are established.(4)After considering various influencing factors such as video signal processing methods,different physiological parameter estimation methods,and different lighting environments,three experiments are designed to verify the stability and consistency of the proposed method in this paper.By calculating the error evaluation index between the non-contact detection results and the medical instrument measurement results under different lighting scenarios,it is found that in each scenario,the mean absolute error(|Me|),standard deviation of error(SDe),and root mean square error(RMSE)of the heart rate are all less than 2.20 bpm,and the |Me|,SDe and RMSE of the respiratory rate are all less than2.40 brpm,and all indicators of Sp O2 are less than 1.60%.By analyzing the Bland-Altman diagram of the results of this method and the contact detection results,it is found that there are two points outside the limit for the deviation of heart rate and respiratory rate,and four points for the deviation of Sp O2 are outside the limit.To sum up,the non-contact measurement results in this paper are highly consistent with the contact measurement values.It shows that the vital signs perception system designed in this paper has good stability and universality,and can be used as an auxiliary detection instrument for routine detection of physiological signs in daily life.