Feature Signal Resampling for rPPG-based Remote Cardiac Pulse Measurement with Streaming Video

Bochao Zou, Dongyue Lv, Xiang-wen Lyu

Abstract: Remote photoplethysmography (rPPG) which can realize contactless measurements of cardiac activity has applications in various fields. However, during practical usage with network cameras, non-uniform sampling caused by the unstable frame rate of the imaging equipment and transmission delay variations of the streaming signal may significantly reduce the performance of existing rPPG algorithms. Therefore, this paper proposes a feature signal resampling method with cubic spline interpolation, combined with an adaptive weight fusion to achieve a robust extraction of pulse waves. The evaluation experiment shows that the proposed method improves the results with streaming videos.