A low-cost, camera-based continuous PPG monitoring system using Laplacian pyramid

Continuous vital signs monitoring plays a critical role in proactively managing people׳s health in their daily lives. Thus, more affordable, portable, convenient, and reliable health monitoring systems are highly demanded. In this study, we present a new approach capable of extracting photoplethysmogram (PPG) from facial videos using the Laplacian pyramid based signal amplification technique, with the goals of improving the estimation accuracy and enhancing the robustness to varying environmental and operational conditions. To validate our approach in the day-to-day setting, 10 videos from each of the 10 subjects were recorded under various lighting and illumination conditions at different places, over a period of one month. All the results are compared against the popular PPG extraction technique based on the region of interest (ROI) averaging method, as well as the PPG waveforms extracted with the standard pulse oximeter. It is shown that the proposed approach can generate comparable results with improved SNR and a high level of consistency and stability. The average accuracy of the proposed approach for heart rate estimation reaches 93.5%, in comparison with 87.9% for the ROI averaging method.