DEVELOPMENT OF AN EFFICIENT ALGORITHM FOR HEART RATE ESTIMATION USING REMOTE PHOTOPLETHYSMOGRAPHY SIGNAL

Abstract

Accurate and non-contact heart rate monitoring has received substantial interest due to its usefulness in telemedicine, fitness tracking, and real-time health surveillance. Traditional contact-based methods, though reliable, pose challenges in terms of user ease, sensor placement, feasibility for remote monitoring. In light of these restrictions, this research explores a remote and scalable solution through Remote Photoplethysmography (rPPG)—a technique that estimates heart rate from face video recorded by a standard RGB camera. The proposed method employs MediaPipe facial landmark detection to identify the facial region which are examples of stable Regions of Interest on the face that are known to have trustworthy blood volume pulse (BVP) signals. The collected RGB signals are converted into the HSV color space to increase the signal extractions’s resilience, where the Saturation (S) channel is isolated due to its reduced sensitivity to illumination variations. This enhances the quality of the extracted signal. Further, a Fourier Decomposition Method (FDM) is applied to isolate heartbeat-related frequency components while reducing the impact of noise and motion artifacts. Experimental validation on publicly available UBFC-rPPG dataset demonstrates that the method performs competitively compared to existing rPPG techniques. The findings confirm that the proposed pipeline achieves consistent and reliable heart rate estimation even under realistic conditions involving movement and lighting changes. This scientific exploration brings forth the growing body of non-contact physiological monitoring techniques, offering a non-intrusive and efficient alternative to traditional heart rate measurement tools. Future work may include enhancing motion resilience, evaluating performance across varying skin tones, integrating deep learning for automatic ROI selection and denoising, and extending the system to measure additional physiological parameters such as respiratory rate. The method holds promise for deployment in consumer health applications, mobile platforms, and large- scale telehealth systems.