Remote Optical Cardiopulmonary Signal Extraction With Noise Artifact Removal, Multiple Subject Detection & Long-Distance

Abstract

Most existing imaging technologies are prone to noise artifact resulting from movement of subjects, movement of the camera, facial expressions, talking, skin tone, and lighting variations. In addition, these technologies have been limited to detecting physiological signs from a single subject at a time. Still, another challenge facing these technologies is that they have only been demonstrated over a limited range to a subject. This paper proposes an efficient remote imaging system with a new, low-complexity method for noise artifact removal to extract the cardiopulmonary signal from a number of subjects (up to six people) and at long-distances (up to 60 m). The experimental results of the proposed system showed a strong agreement, high correlation, and low noise level with the reference measurements and outperformed the conventional measurement methods such as independent component analysis and principle component analysis in both the stationary and non-stationary scenarios.

Keywords

• Imaging photoplethysmography (iPPG)
• cardiopulmonary signal
• video magnification
• signal decomposition technique
• blind source separation (BSS)
• canonical correlation analysis (CCA)