Research On Accurate Extraction Algorithm For Visual Photocapacitance Product Pulse Wave Signal

The pulse wave signal contains many vital sign information,which is of great significance for daily health monitoring and disease diagnosis and treatment.However,contact based photocapacitive pulse wave measurement equipment requires contact with the detector’s skin,which is not friendly for measuring vital signs of severely burned patients and newborns.Therefore,non-contact pulse wave measurement technology based on visual optoelectronic volume has emerged.This article uses visual optoelectronic volume technology to achieve accurate extraction of pulse wave signals.In order to convert the two-dimensional image of the video frame by frame into a temporal waveform,it is necessary to perform initial extraction of the pulse wave signal.The initial extraction process uses the first layer of the Laplace pyramid to separate the human body part from the background part of the image;Adopting inter frame windowing difference method for selecting regions of interest,thereby greatly eliminating background noise;Adaptively select the amplification parameter with the highest energy ratio for video motion amplification,which weakens the synchronous amplification of noise while amplifying the small pulse motion,and improves the signal-to-noise ratio.In the signal denoising section,the segmented baseline elimination method is used to maintain the pulse wave signal at the same baseline level for baseline drift caused by environmental noise;To eliminate the environmental noise contained in the human body,a modulation domain spectral subtraction method is used to effectively enhance the pure pulse wave signal at the "time frequency" level.In the decomposition and reconstruction part of the signal,the parameter corresponding to the maximum Pearson correlation coefficient is adaptively selected to reconstruct the pulse wave signal,removing the clutter present in the pulse wave while preserving the pulse wave feature points as much as possible.In the experimental part,a non-contact pulse wave measurement device is first built,followed by system performance testing and hardware parameter selection experiments to determine the optimal working conditions.Finally,internal algorithm parameter selection experiments and algorithm accuracy and repeatability experiments are conducted.In the accuracy experiment,the average absolute error of heart rate measurement is 0.36 BPM,which is higher than the results of the contact pulse wave measurement instrument;The standard deviation of error is 0.27 BPM,indicating a concentrated error distribution and strong predictability;The Pearson correlation coefficient is 0.67,which has a good matching effect and can well reflect the physiological characteristics of pulse waves.In the repetitive experiment,the mean heart rate measurements of 30 subjects were calculated using the scheme proposed in this paper and measured with a contact pulse wave measuring instrument,resulting in 69.20 BPM and 69.17 BPM,respectively.The consistency estimation chart using the Bland Altman method shows good consistency between the two.In 100 repeated experiments on a subject,the standard deviation of heart rate was 0.67 BPM,and all heart rates were within the confidence rate range of 99.73%,indicating that the proposed protocol has high repeatability and stability.