Research On Non-contact Human Heart Rate Detection Method Based On Face Video

In recent years,with the rapid development of our society,people’s pace of life is getting faster and faster.Irregular diet,high workload,lack of physical exercise lead to cardiovascular diseases have a more significant impact on the health of residents,and the incidence of cardiovascular diseases continues to increase.Heart rate is an important physiological index that indicates the state of the human body.Doctors can evaluate the body of patients with cardiovascular diseases through heart rate,and healthy people can also prevent the occurrence of cardiovascular diseases through heart rate detection.The video heart rate detection method based on ballistocardiography(BCG)has achieved relatively ideal experimental results,but in the real scene,due to the interference of environmental noise,such as environmental illumination change,motion interference and other factors,the experimental results have a great impact.Therefore,it is of theoretical and practical significance to study how to remove the influence of environmental noise to accurately detect human heart rate in video.Aiming at the problem of noise interference,this paper has done the following work:(1)The disadvantages of BCG signal captured by motion tracking being susceptible to light illumination change and motion interference were analyzed,and the Steerable filters schematic was selected to extract BCG signal.The common BCG signal denoising algorithms are introduced.(2)To solve the problem that the change of ambient illumination has influence on the extraction of heart rate information,a video heart rate detection method based on Steerable filters schematic was proposed.Firstly,face detection is carried out on the input video,and the target detection area is selected by the box.Secondly,a vibration signal extraction method based on Steerable filters schematic was designed to extract the phase difference signal of the target region,and for the phase difference signal,bandpass filtering and principal component analysis were carried out to filter out the interference noise and establish BCG signal.Finally,the BCG signal is transformed into the frequency domain by Fourier transform,and the heart rate value is extracted from the spectral density function.Experimental results show that this method can adapt to different illumination environments,and the accuracy of heart rate detection is better than the existing motion tracking capture BCG signal extraction methods.(3)A video heart rate detection method based on Variable Mode Decomposition(VMD)and Low-rank Sparse Matrix Decomposition(LSMD)was proposed to solve the motion interference problem caused by facial expression changes.The method is based on the extraction of the head phase difference signals through directional controlled filter,and then the extraction of the eigenmodes containing the heart rate information through VMD.Finally,the obtained signal matrix is decomposed by LSMD,and the movement interference is removed to obtain the pure heart rate signal.Experimental results show that this method can effectively remove the motion interference caused by facial expressions.