Automatic Dual-Exposure for Illumination-Robust Remote Physiological Monitoring

Camera-based vital signs monitoring is a noncontact and convenient method for extracting physiological signals from human face and body. However, current methods struggle with varying lighting conditions, such as inhomogeneous or uneven illumination. This study introduces a novel Automatic Dual-Exposure (ADE) method that adjusts two different exposure settings sequentially for both the face and chest in real-time, optimizing facial brightness for heart rate (HR) measurement and chest brightness for respiratory rate (RR) measurement. Experiments with 14 participants demonstrate that ADE significantly outperforms the fixed global exposure method of an industrial camera, improving measurement accuracy from 77.56% to 95.99% for HR and from 66.07% to 95.58% for RR. This shows the robustness of the ADE strategy in dynamic and uneven illumination conditions.