Color signal processing methods for webcam-based heart rate evaluation

Abstract

Computer vision methods are widely applied in health assistance and medical diagnostics. Photoplethysmography (PPG) is one such method that can be used for contactless estimation of heart rate through the analysis of slight variations of skin color which are caused by changes in the blood volume in vessels. These changes of skin color registered by a camera are called color signal. According to recent studies some PPG methods can be applied on video data recorded by common web-cameras with sufficient accuracy, so they are recognized as potentially applicable for long-term health monitoring in house or office conditions. In this study, we evaluate the accuracy of commonly used signal processing methods for webcam-based PPG as well as novel modifications of these methods in various combinations with preprocessing and postprocessing filtering algorithms. In particular, the Extended Fourier analysis that is based on Gaussian smoothing and temporal averaging of Fourier spectra was applied to estimate heart rate.