Noise-resilient heart rate detection in automotive environments using orthogonal projection

This paper addresses the challenges of heart rate (HR) measurement using remote photoplethysmography (rPPG) in automotive driving environments, where rapid lighting changes, head movements, and vehicle vibrations introduce significant noise interference. We propose a novel pulse extraction method based on coordinate delay transformation, which applies time-delay processing to each region of interest (ROI) signal, generating a multi-channel ROI signal set. Signal decomposition is then utilized to separate pulse and noise components, and an orthogonal projection is employed to remove noise interference from the rPPG signal. The proposed method is validated on a self-collected dataset and compared with three state-of-the-art near-infrared rPPG algorithms. Experimental results demonstrate that the proposed approach excels in HR detection in driving environments, effectively handling noise disturbances and providing long-term, stable HR measurements.