Human heart rate monitoring based on facial video processing

Abstract

Heart rate (HR) is one of the most important physiological parameters and a vital indicator of people’s physiological state, making it important to monitor. Over the last decade, research has focused on non-contact systems, which are simple, low-cost, and comfortable to use. This paper analyses the complexity of each step in the development of a human heart rate monitoring algorithm based on facial video processing. Specifically, the research focuses on the pulse signal extraction step. The proposed algorithm based on the transform of 2D signal to 1D signal, its detrending and window discrete transform are used to improve the accuracy of HR estimation. The experimental results show that the accuracy of human heart rate estimation in terms of MAE and RMSE is equal around 2 bpm.