Research On The Application Of KCF Algorithm In Video-based Non-contact Heart Rate Measurement

As one of the important parameters to evaluate human health,heart rate can effectively reflect human health status.Therefore,heart rate monitoring plays an irreplaceable role in human health monitoring,especially in wards and gyms.At present,the heart rate monitoring system mainly use the method of body contact,such as bracelet and chestbelt,which brings certain influence to the tester during the process of measurement,for its complicated operation and need contact human body for a long time,the heart rate monitoring system brings certain influence to the tester.However,most of the non-contact measurements in the market are based on mobile phones and radars,which are insufficient to detect the human body in motion.In this study,the heart rate monitoring system based on KCF algorithm was studied.Through KCF target tracking,video information containing face was processed,heart rate signals in the video were extracted,and heart rate values were calculated.This method was realized by raspberry PI.This method can realize continuous non-contact heart rate monitoring without changing the user’s daily habits,and has the characteristics of antimovement interference,comfort,easy to operate and easy to carry.The implementation of this method is of great significance to the detection of fitness staff’s physical state in the gym.The main work of this paper is as follows:1.compare various face detection and tracking algorithms,demonstrate the importance of face detection and tracker performance for signal extraction in non-contact heart rate monitoring,and demonstrate the superiority of KCF target tracking algorithm in its signal extraction.2.Non-contact heart rate monitoring scheme is designed based on KCF target tracking algorithm.The accuracy and rapidity of KCF algorithm are used to reduce the influence of human movement on the accuracy of non-contact heart rate measurement.3.The KCF algorithm is optimized and improved to accelerate the running speed,so the algorithm can better serve the real time requirement of non-contact heart rate measurement.4.Consistency assessment.The bland-altman diagram and Pearson correlation coefficient were used to evaluate the consistency between the non-contact heart rate measurement method and the chest band measurement data.5.The microcomputer system is used to realize the non-contact heart rate monitoring system.With the help of the raspberry PI microcomputer system,the above non-contact heart rate monitoring system is implemented in the raspberry PI and the product is realized.Experimental results show that the static state,slight motions and strenuous exercise state,the heart rate monitoring system based on KCF algorithm with chest belt type contact measurement results with high consistency,can satisfy the demand of heart rate monitoring in the process of exercise,and put forward the non-contact heart rate monitoring raspberries pie hardware implementation,meet the needs of the non-contact measurement of the transition.