Research On Non-contact Heart Rate Detection Method Based On Ordinary Color Camera

Heart rate is an important indicator that can effectively reflect the status of the heart.Traditional heart rate detecting methods require close contact with the skin of the body surface,which brings discomfort to the subject.Recent studies found that imaging plethysmography can detect the color changes of human faces through ordinary color cameras to achieve heart rate detection.This technology has the advantages of non-contact and low cost,but the measurement results are easily affected by motion interference.Besides,the main interference factors for different IPPG devices are also different.When using a common camera for face video collection,the measurement results are easily affected by facial expression changes.When a smartphone is used as the heart rate detection tool,hand shaking becomes the main factor to reduce the detection accuracy.In order to improve the accuracy of heart rate detection of IPPG devices,the following work has been done in this study in response to the above problems:(1)In order to address the noise problem in extracting heart rates by ordinary cameras,a heart rate detection method against facial expression changes is proposed.This meth od first divides the human face into multiple sub-modules and then constructs a pulse si gnal matrix.Since the ideal pulse signal matrix has low rank characteristics,an adaptive matrix separation algorithm is used to remove the motion noise in the matrix;from whi ch sub-modules with rich heart rate information are selected to establish the fused pulse signal.Finally,this approach calculates the power spectral density of the pulse fusion si gnal and achieves the heart rate estimation.(2)Aiming at the problem of jitter interference in the extraction of heart rate by smartphones,an anti-jitter heart rate detection method is proposed for mobile phone.This method first removes baseline drift and sharp noises in the pulse signal by baseline cancellation,spike smoothing and five-point cubic smoothing algorithms;then,taking advantages of the short-term stationary characteristics of the pulse signal and using the delayed signal as the desired signal of an adaptive filter to eliminate noise and restore signal details;finally,the clean pulse wave is obtained and heart rate could be estimated on the basis.In order to verify the performance of the proposed method,a large number of face videos were collected through ordinary cameras and smartphones.30 testers participated in the experiment with 351 video samples recorded.Experimental results show that the method proposed in this paper is superior to the classic heart rate detection algorithms in various test environments,and can effectively remove interference noise and improve the robustness of heart rate detection.