Reveal Sadness By Separating Facial Blood Active Signal

Physiological arousal,a physiological response,would change heart rate and blood pressure.Meantime,the increasing daily pressure lets people sadder,which would harm people's physical and mental health like causing them suffering from cardiovascular disease,cognitive impairment and mental disorder.Although it is urgent for people to detect the potential sadness,there are many problems in the popular detection ways.On the one hand,the psychometry like questionnaire depends on people's subjective response largely,and affects by sensory factor.On the other hand,the accuracy of biometric likes Electroencephalogram is high,but they need special equipment's and experts' assistance,and are inconvenient.Photoplethysmographic is a new noncontact real-time method which can measure facial bloodstream signal.By recording facial video and analyzing the change of scattered light,it can output facial bloodstream signals such as heart rate,blood pressure and heart rate variability.Through mapping relationship between the 10 eigenvalues of those 3 signals and the psychological state,people's emotional state can be got.Based on this theory,a new remote,hidden and contactless psychology measuring method was constructed,while solved the issues of high subjective,high specialized,inconvenient and discomfort.The main work is as follows:(1)Optimized the heart rate measuring program by Photoplethysmographic.The program can accurately measure the heart rate under the 6 degrees of freedom facial motion and mask by data acquisition in 2 subregions,removing artifacts and multi angle facial recognition.Compared with the results of electronic sphygmomanometer,the accuracy was 97%.(2)Established a blood pressure measurement way via the phase difference of pulse wave in subregions.The pulse wave transit time is deduced by the phase difference of pulse wave propagating in the facial subregion.Output blood pressure by the linear relationship between it and transit time,the accuracy was 94.3%.The relationship between mean pressure and heart rate was established based on Ring Quadrupole.The accuracy was 96.1%.It replenishes the missing data causing by blood pressure minutely output.(3)Analyzed heart rate variability's time-domain features and frequencydomain features.Its time-domain features are extracted,and the frequency-domain features are interpolated and analyzed by different methods.The results are less affected by noise and the peak curve is smoother.(4)Verified the relation between facial bloodstream signal and sadness.Created a Valence – Arousal model and chose Aff-wild2 data set to find 10 eigenvalues for proving an evidence that program can verify sadness in a remote and contactless way.The accuracy is 85%.The results show that: analyzing the changes of facial bloodstream information can bring an objective and professional sadness evaluation method through shooting facial video in a non-invasive and continuous way.