Motion Robust Remote Photoplethysmography via Frequency Domain Motion Artifact Reduction

Remote photoplethysmography (RPPG) is an emerging technology for measuring vital signs. It paves a way for non-contact measurement and as a result can be used in multiple scenarios like telemedicine, fitness centers, preliminary health screening, etc. RPPG techniques extract heart rate from the video recording of the subject. The challenge in this field is to extract the heart rate when the subject is moving. The present work proposes frequency domain motion artifact reduction (FDMAR), an effective method to filter the motion artifact for better heart rate measurement. In this method, we select a point that moves in sync with ROI and then use this point to cancel out the motion artifact. The experiments were performed under the white light illumination and subjects were free to perform natural movement while recording the videos for heart rate extraction. Results have shown significant improvement in the spectrum by suppressing the motion artifact noise. With this method, we could achieve +/- 4 bpm for moving subjects and +/- 2 bpm for still subjects.