Motion Robust Imaging Ballistocardiography Through a Two-Step Canonical Correlation Analysis

Imaging ballistocardiography (iBCG) is a video-based noncontact technique to detect heart rate (HR) from weak mechanical head movements caused by heart beating. However, rigid motions caused by voluntary movements and nonrigid motion resulted from facial expressions can easily distort the iBCG measurements. In this article, we propose a novel method, called robust iBCG (RiBCG), to suppress motion artifacts in iBCG with a two-step canonical correlation analysis (CCA). First, feature points are determined and tracked within two regions of interest (ROIs) from the face, where the vertical traces are taken as raw iBCG signals. Next, the first CCA is taken to separately remove the shared rigid motion artifacts between the horizontal and vertical traces in each ROI, where the obtained rigid-motion-free iBCG signals are further compressed by principal component analysis (PCA). Then, CCA is applied again to two sets of principal components to suppress nonrigid motion artifacts with low spatial correlations. Finally, the target HR value is determined as the one with the highest peak of power spectrums among all canonical variates (CVs). Besides, an improved version of RiBCG, termed RiBCG-C, is also proposed to reduce the HR outliers considering the continuity of HR variations. The proposed methods, as well as several other typical video-based HR measurement methods, are evaluated on two public databases, UBFC-RPPG and COHFACE, where the proposed RiBCG-C method achieves overall the best performance. The study provides a promising scheme for RiBCC measurements under realistic application scenarios.

Automated summary

This article proposes a novel method called robust iBCG (RiBCG) to suppress motion artifacts and an improved version RiBCG-C to reduce HR outliers. Through evaluation on public databases, RiBCG-C method achieves overall the best performance, providing a promising scheme for RiBCG measurements in realistic application scenarios.