Remote Photoplethysmography Heart Rate Variability Detection Using Signal to Noise Ratio Bandpass Filtering

Robust Heart Rate Variability Measurement from Facial Videos

“…In recent years, there has been a growing interest in heart rate variability (HRV) estimation using remote photoplethysmography (rPPG), and many researchers have focused on developing robust and accurate algorithms for this purpose. Typically, a pipeline for rPPG-based HRV estimation includes several stages, such as face detection and tracking, skin segmentation, region of interest (ROI) selection, and rPPG construction [2][3][4][5]. In addition, there are numerous post-processing steps that can be applied to clean, filter, or denoise the rPPG signal to improve the accuracy of HRV estimation.…”

“…A wavelet-based approach was proposed by Huang et al [3] and He et al [4]. Huang et al [3] sourced the rPPG signal by utilizing the CHROM method [10] and further added a post-processing step based on a continuous wavelet transform, termed CWT-BP and CWT-MAX.…”

“…During the CWT-MAX step, windows from the signal are chosen and coefficients that have the largest values within a particular window are selected to reconstruct the signal by inverse CWT. He et al [4] further improved CWT-based denoising methods by introducing CWT-SNR, which selects coefficients based on the signal-to-noise ratio of the reconstructed rPPG signal. Both methods implemented peak-detection algorithms to acquire time-domain HRV metrics and heart rate.…”