In this paper, we propose the method of remote estimation for heart rate (HR) and heart rate variability spectrogram (HRVS) by analyzing hemoglobin concentration from facial RGB images in low light environment. The emotion monitoring has a great potential in areas such as market research, safety measure, medical, and robot systems and so on. Various methods were proposed for the remote emotion estimation. Especially, the methods to estimate emotion by analyzing physiological signals from RGB video are expected to be used practically. However, these studies could not be applied in dark places where monitoring is necessary such as infant's bedrooms, crime prone roads and car drivers. Therefore, the proposed method uses the high sensitivity RGB camera capable of capturing videos at low illuminance. As the result, we could measure HR with over 99% accuracy and estimate HRVS with high accuracy in low light environments of 10 lux that is the same brightness as the twilight where this low light environment is required in monitoring infant's bedrooms, crime prone roads, car drivers.
Genki Okada, Keiichiro Kagawa, Shoji Kawahito, Norimichi Tsumura, "Noncontact heart rate measurement using high sensitivity camera in low light environment" in Proc. IS&T 25th Color and Imaging Conf., 2017, pp 159 - 164, https://doi.org/10.2352/ISSN.2169-2629.2017.25.159