Research And Realization Of Heart Rate Detection Based On Face Video Image

In the normalized physiological monitoring of the patient,the heart rate is one of the vital signs that the doctor must first detect.With the rapid development of society and the expansion of urban population,doctors are increasingly scarce resources.Each doctor needs to take care of more and more patients,so there is an urgent need for telemedicine.Aiming at the limitation of the use of existing heart rate detection techniques,such as physical contact and space constraints,this paper designed and realized a non-contact remote heart rate detection system by using the face video images and PhotoPlethysmoGraphy(PPG)principle.This project is based on the project of the Chinese Academy of Sciences(part of the "863" project of the Chinese Academy of Sciences).This system acquis face image by raspberry pie ARM platform,and realizes heart rate detection with the images remotely.The main research work of this paper is as follows:(1)This Paper designed the raspberry pie collection programmer firstly,and then completed the face image acquisition.The face detection was realized quickly and accurately based on the Haar feature and the Viola-Jones algorithm.The zoom and intercept operations were performed according to the distance and deflection angle of the feature points corresponding to the two eyes.The optimal measurement area segmentation was completed based on the target region tracking algorithm.(2)This paper summarized the algorithm of extracting heart rate information from face image signal which were based on PPG principle.The heart rate detection algorithm was designed in detail based on FastICA which included RGB three-color separation,independent component analysis and spectrum analysis.The noise reason of PPG heart rate signal and the existing de-noising methods were analyzed.This paper compared the related experiments by using the single channel ICA algorithm and other algorithm.(3)The realization of heart rate detection system.A remote management and human-computer interaction module was designed which realized communication,display,control,calculation and other functions that is based on the embedded device of raspberry pie.Finally,the heart rate detection system designed by this topic could remotely detect the heart rate information of the subjects through the face image.The difficulty and innovation of this paper were as follows:(1)Aiming at the influence of motion artifact on PPG signal,This paper proposed a tracking method based on eye location.The region edge was weighted by twodimensional Gauss distribution to reduce the noise interference introduced by motion artifact.(2)In view of the same frequency aliasing noise in PPG signals after blind source separation,a dynamic channel embedding method was proposed to analyze the green channel source signals using single channel ICA correlation algorithm to minimize the noise interference.(3)In the image acquisition module of heart rate detection system,the problem of loss of PPG information was caused by the low frame rate of PPG signal acquisition in the experiment.The image acquisition scheme was researched and improved based on multi thread and IO multiplexing technology to guarantee the integrity of PPG signal in face image.