| With the aging of China's population,the high incidence of cardiovascular and cerebrovascular diseases in the elderly has become a problem that people need to face.Arteriosclerosis,as one of the main causes of cardiovascular and cerebrovascular diseases,has become a major difficulty in the prevention and treatment of cardiovascular and cerebrovascular diseases due to it is slow occurrence and irreversible damage.At present,most of the arteriosclerosis detection equipment are expensive professional medical equipment,not portable,and the existing diagnostic equipment is almost all contact-type,not comfortable,nor suitable for 24-hour monitoring of arteriosclerosis,making arteriosclerosis The degree of early self-examination is very inconvenient.Based on the pulse wave velocity method and the photoplethysmography method,this paper abandons the contact sensor used in traditional arteriosclerosis measurement methods,collects noncontact video signals of the body's surface photoelectric volume through a mobile phone camera to detect the changes in light intensity absorbed and reflected by the tissues such as blood and muscle to extract the blood volume pulse wave information,and the calculate the pulse wave velocity to realize contactless estimation of the degree of arteriosclerosis.This low-cost,easy-to-operate routine assessment method is of great significance for the prevention and treatment of cardiovascular disease.The main contents of this article are as follows:(1)In order to eliminate the interference and motion artifacts caused by changes in shooting distance and slight shaking,an algorithm for locating the forehead area using the relative position of the face was proposed during the extraction of the facial region of interest.Based on the overall facial recognition of the HOG algorithm,the feature points and their relative positions are used to accurately locate the forehead area.(2)The palm recognition algorithm is added for the acquisition of the hand pulse wave signal,which is abandon the traditional finger clip sensor,so that there is no contact during the entire detection process.Use the ssd-mobilenet network to train a hand recognition classifier on the server,use this classifier to accurately detect the hand area,and propose an algorithm for determining the palm area based on the boundary distance calculating by mask and contour detection.The palm area is accurately segmented from the hand area to improve the accuracy of the pulse wave signal.(3)In the process of extracting the pulse wave signal,the FastICA algorithm and FIR filtering and the gradient maximum are used to locate the feature points.Blind source separation uses FastICA instead of the traditional ICA algorithm,which improves the convergence speed during blind source separation,reduces the computational burden of computing equipment,and uses G-channel color components to perform correlation analysis on independent signals after blind source separation to solve disordered sequence,select the pulse wave independent source signal.The FIR filter is used to solve the phase drift phenomenon that occurs in the past filters,so that the characteristic points of the pulse wave correspond to the time more accurately.Using the characteristics of equal sampling intervals,the feature point of the first derivative maximum in the discrete pulse wave signal is located by calculating the gradient maximum value,which avoids the situation of missing information caused by the fitting curve during the derivation process.(4)The pulse wave velocity is calculated using the pulse wave signals from each region of interest,and the consistency of this method and the ECG-PPG combination is explored using the Bland-Altman analysis.The analysis results prove that the two have a good degree of consistency,that is,the method can meet the needs of daily arteriosclerosis assessment. |
PDF Full Text Request