Non-contact Heart Rate Measurement Algorithm Based On Skin Detection

Heart rate is an important physiological parameter that reflects the health of the human heart and arteries.The long-term monitoring of heart rate can effectively control cardiovascular disease.Traditional heart rate measurement contacting with people is less suitable for skin sensitive groups,and its measurement experience of touch and squeeze is not conducive to long-term monitoring.The cost for non-contact heart rate measurement by microwave or laser can be high,which may harm health of users.Therefore,non-contact heart rate measurement based on camera has become one of the hotspots of current research.Its convenience,safety and low cost have high application value in telemedicine and intensive careThe systolic and diastolic motion of the heart can cause periodic changes in blood volume of blood vessels,so the intensity of diffuse reflection on skin surface changes periodically under visible light.Based on the principle,the research content and results for the non-contact heart rate measurement algorithm based on video image are as follows(1)Based on the optical characteristics of skin,skin detection algorithms by color and principle for non-contact heart rate measurement are analyzed.The distribution range and characteristics of skin color in different color spaces are studied,which lays a foundation for the design of skin detection algorithm.At the same time,combined with the principle of PhotoPlethysmoGraphy(PPG),the interaction model between skin and light is deeply analyzed,and the theoretical basis of non-contact heart rate measurement is studied.The principle of heart rate measurement based on skin optical properties is visually validated by Eulerian Video Magnification algorithm(2)For the extraction of physiological signals,a specific scheme of heart rate prediction based on skin video image is designed.PPG signals containing pulse information are analyzed by red,green and blue original one-dimensional chromaticity signals separated from regions of interest.This paper presents an improved algorithm for extracting PPG signal based on Blind Source Separation(BSS),which focuses on the randomness of component selection and improves the stability and accuracy of heart rate prediction.At the same time,this paper proposes a PPG signal extraction algorithm based on chroma model.By further analyzing the skin optical model,the effect of melanin on light absorption is reduced and the robustness of the signal is improved.(3)A non-contact heart rate measurement system based on camera is built.Combined with skin detection algorithm and two PPG signal extraction algorithms designed in this study,the average of heart rate is calculated as the output of the system.The system is implemented on hardware by a color industrial camera with resolution of more than 1080p and PC,and its software is realized by C++and MATLAB mixed programming.(4)Through Bland-Altman analysis,the consistency between the predicted heart rate of the system and the reference heart rate and the consistency of the heart rate obtained by the two improved non-contact algorithms are verified.Without strict control of illumination and the state of the subjects,the average relative error of heart rate prediction is 3.46%,which has high stability and robustness.The effects of different body parts,illumination and frame rate on non-contact heart rate prediction are analyzed through comparative experiments.Experiments show that strong signal can be extracted from face and palms,and the signal-to-noise ratio(SNR)of the signal can be improved by uniform and bright light and high frame rate.