Detail-preserving arterial pulse wave measurement based Biorthogonal wavelet decomposition from remote RGB observations

Abstract

Extracting arterial pulse wave from the video, also known as Imaging Photoplethysmography (IPPG), has emerged as an efficient health monitoring approach. The detail-preserving arterial pulse wave can significantly improve healthcare efficiency through the provision of physiological and pathological information. However, the accurate measurement of arterial pulse wave by IPPG is still challenged by the fact that the pulsation is weakly expressed at the skin surface and extremely vulnerable to interferences. This work proposes a novel approach to measure the detail-preserving arterial pulse wave via interference isolation from facial video. A skin reflection model is introduced to guide the design of the IPPG measurement system, where the IPPG signal is pre-filtered to obtain the variation related to the pulse wave, and a Biorthogonal wavelet is further employed to distinguish the unique pattern of arterial subtle variation from interferences. Experimental results validate the efficiency of proposed method with a correlation coefficient of 0.92.