Vision-based heart-rate monitoring is an innovative technology used in recent health care equipment development. Most of the current solutions involve the use of visible light, which are feasible in well-controlled indoor lighting environment but not at nighttime or outdoor settings. Recently, the use of vision-based heart-rate monitoring in a variety of real-life settings, such as sports training, driving fatigue detection, anti-spoofing detection in Fin-Tech application and emotion recognition, has been explored. A robust detection technology that can handle adverse conditions such as poor lighting will match the needs of users in those applications. This project aims to provide a real-time heart rate detection solution in various background lighting conditions. Based on the infrared light’s optical absorbance characteristics for oxyhemoglobin and deoxyhemoglobin, we can measure the variation of hemoglobin oxidation status across time. Heart rate can then be deduced by the variation of hemoglobin oxidation. In this project, we aim to explore a new approach using dual wavelength infrared light to improve the signal to noise ratio from NIR videos by comparing the signals from two infrared images with different wavelengths. We plan to test the accuracy of this approach in environments with different levels of brightness and color temperature. We will also explore the possibility of detecting multiple persons’ heart-rates simultaneously. The results of this project will provide solutions for various applications, which include physiological monitoring for sports training, health fitness monitoring for senior, detection of driving fatigue for Advanced Driver Assistance Systems (ADAS), sleeping quality measurement and so on. We will also build a foundation for future non-contact based emotion recognition technology for smuggler detection conducted by customs officers using computer vision. In long term, this technology can be applied in various health- and medical-related equipment to promote public health and safety.
Visual-based Heart-rate Monitoring (ARD/252)
01 / 05 / 2020 - 31 / 01 / 2021
Mr Joe Tak-Fu LAU