Research On Heart Rate Detection Method Via Face Video Processing

Heart rate is one of the important indicators that can directly reflect the health of the human body.Heart rate detection is applied to many aspects of the medical field,such as physical examination,major surgery,and postoperative treatment.Heart rate detection based on face video processing has recently been performed through a noncontact manner without complex operations and sense of restraint.However,the existing methods cannot predict well in complex realistic scenes including interferences of shaking target,illumination conditions,and expression changes,which lead to the inaccurate blood volume pulse(BVP)and unsatisfactory detection accuracy.Interference factors are different with different people.Face shaking and unconscious expression changes are relatively weak for ordinary people,but extremely easy and severe for babies.In order to eliminate the interferences in video and solve the difficulties of population applicability,this paper makes a thorough study on tracking method for shaking face and extraction model of clean BVP signal.Two novel heart rate detection methods based on face video processing are proposed,which are suitable for ordinary people and babies respectively.The innovations are summarized as follows.(1)A novel heart rate detection method is proposed to eliminate the interference of face shaking based on face video processin.Firstly,this method combines feature point detection with image slant correction to track shaking face and weaken the interference of face shaking.Secondly,reasonable spatial scale characteristics and frequency ranges,which reflect the heartbeat,are selected to amplify the skin color variation of face video and extract the clean BVP signal.Finally,it overcomes the weakness of small samples in heart rate estimation process by iterative Fourier coefficient interpolation.(2)A novel heart rate detection method for babies based on face video processing is proposed.This method focuses on the expression interference in baby videos.After eliminating the face shaking of video,some image blocks where the effect of facial expression changes is weak are adaptively selected to calculate the BVP signal by using face multi-region segmentation and weighted fusion mechanism.It can greatly eliminate the facial expression interference.Then,the wavelet decomposition is introduced to remove the baseline wander noise,thus cleaner BVP signal is acquired for heart rate estimation.(3)The analysis of heartbeat regularity based on face video is explored.The consistency between electrocardiogram and the video BVP wave is verified via matching the BVP wave extracted from face video with electrocardiogram captured synchronously.On this basis,the R-R interval of pulse wave is detected to analyze heartbeat regularity of testers.Using this method,arrhythmia can be preliminarily predicted.To validate the better performance of the methods proposed in this paper,many ordinary people and baby videos are collected for experiments.The results show that our methods have higher accuracy and stronger robustness compared with the existing video-based heart rate detection methods.Besides,the analysis of heartbeat regularity based on face video proposed in this paper provides feasible suggestions on contactless prediction for some heart diseases like arrhythmia,etc.