Non-Contact Vital Signs Measurement Based On Human Face

Search and rescue robots are playing an increasingly important role in wars,natural disasters,nuclear bio-terrorist incidents and public health and security incidents.Robots can replace rescue workers in the complex and unknown dynamic environment to carry out the search and rescue of the wounded(or survivors),which will help improve the rescue efficiency.For the search and rescue robots,the ability to quickly and accurately perceive the physiological parameters of the wounded has become one of the key characteristics of intelligence.However,the traditional vital signs measurement method e.g.manual and wearable is hard to adapt to the future robot autonomous search and rescue needs.Researching and developing the front-end autonomous ability to access the vital signs of the wounded timely,for achieve the active identification of the wounded,which it's important to improve the unmanned intelligence of rescue robots.This paper takes unmanned search and rescue robot as the target carrier platform for application and exploration,and combines with the characteristics of real search and rescue operations,carries out research on the measurement of non-contact vital signs based on human face.By depth profiling Remote Photo Plethysmography(RPPG)technology biological optical principle,the paper points out that the advantages and characteristics of face as physiological signal acquisition region of interest(ROI),based on the basic technical framework in the integrated application of face detection and tracking methods,maximal ratio combining(MRC)algorithm,Synchrosqueezing Transform(SST)methods to enhance RPPG technology robustness,accuracy and adaptability in unmanned platforms,and then builds simulation experiments to evaluate the real performance of the proposed method.This paper analyzes the basic scientific theory of RPPG technology,and introduces the formation mechanism and characteristic points of the human systemic circulatory system and Blood Volume Pulse(BVP)from the physiological basis.The optical absorption principle of skin tissue was revealed,and the advantages and characteristics of human face as ROI were discussed.The basic flow framework of RPPG technology is formed and its influencing factors and evaluation indexes are expounded.In view of the difficulties and challenges faced by RPPG technology in the application of unmanned platform,this paper improves and optimizes the key technical links to improve the overall performance of the technical system.Viola-Jones face detector is used to automatically initialize the face area in the video images,and Kanade-Lucas-Tomasi(KLT)corner feature tracking method is used to track the face region,so as to improve the extraction efficiency of the face in the process of movement and enhance the value of the PPG signal extraction region.The MRC algorithm is used to combine the PPG signal in the pixel area of the face image to improve the signal-noise ratio(SNR),so as to obtain the robust PPG signal,and then the heart rate was extracted.A variety of experimental scenarios were designed,such as stationary face,head moving and talk to verify the application performance of the proposed method.In this paper,three kinds of respiratory-induced variation modulation are introduced,and the formation mechanism and the influence of different modes on PPG are analyzed.SST method was used to reconstruct the amplitude,frequency and intensity components induced by respiration on PPG signal,after obtaining the frequency-time spectrum of each component,the accurate and stable instantaneous respiratory frequency was detected by fusing the three frequency peaks.The respiratory rate extraction method was evaluated in the benchmark dataset.Verification of heart rate and respiration rate measurement algorithm and preliminary construction of embedded prototype were carried out in complex application scenarios.Different positions of the wounded were simulated by the positive side faces of two lying postures,to test the application performance of the proposed algorithm in Unmanned ground vehicle(UGV).The vital signs measurement performance of UGV and unmanned aerial vehicle(UAV)in different face acquisition distances is also compared to highlight the adaptability of unmanned platform.Embedded intelligence computing device is used to explore the application of the proposed algorithm preliminary.This paper innovates the application of unmanned platform as the technical target carrier,and proposes the application of RPPG technology to the acquisition of physiological parameters of the wounded for the first time.Combined with the platform and technical advantages,the MRC algorithm is innovatively used to merge the diversity of multiple signal paths to obtain the maximum SNR,so as to obtain the robust PPG signal for the measurement of heart rate and respiratory rate.The effects of three kinds of respiratory-induced variation modulation of PPG signal were fused.This paper designs innovation experiments to verify the proposed method of real application performance,posture,distance measuring distance,in a variety of common space under the experimental conditions,such as unmanned platform to carry out the experiment,the results show that this method compared with the "gold standard" method and benchmark comparison algorithm,the method to realize non-contact heart rate,breathing rate,high measurement accuracy,and good robustness and good adaptability to applications in different platforms.The research results of this paper have achieved preliminary results in improving the remote vital signs acquisition ability of unmanned platforms,providing important ideas for the construction of simulation experiments for the wounded,and laying a foundation for further improving the actual application performance of search and rescue robots in the implementation of rescue for the wounded.