Search

Blood pressure (BP) is one of important vital signs in diagnosing certain cardiovascular diseases such as hypertension. A few studies have shown that BP can be estimated by pulse transit time (PTT) derived by calculating the time difference between two photoplethysmography (PPG) measurements, which requires a set of body-worn sensors attached to the skin. Recently, remote photoplethysmography (rPPG) has been proposed as an alternative to contactless monitoring. In this paper, we propose a novel contactless framework to estimate BP based on PTT. We develop an algorithm to adaptively select reliable local rPPG pairs, which can remove the rPPG pairs having poor quality. To further improve the PTT estimation, an adaptive Gaussian model is developed to refine the shape of rPPG by analyzing the essential characteristics of rPPG. The adjusted PTT is computed from the refined rPPG signal to estimate BP. The proposed framework is validated using the video sequences captured by an RGB camera, with the ground truth BP measured using a BP monitor. Experiments on the videos collected in laboratory have shown that the proposed framework is capable of estimating BP, with a statistically compliance compared with BP monitor.

Recently, there are many works on discriminant analysis, which promote the robustness of models against outliers by using L1- or L2,1-norm as the distance metric. However, both of their robustness and discriminant power are limited. In this article, we present a new robust discriminant subspace (RDS) learning method for feature extraction, with an objective function formulated in a different form. To guarantee the subspace to be robust and discriminative, we measure the within-class distances based on $\text{L}_{2,s}$ -norm and use $\text{L}_{2,p}$ -norm to measure the between-class distances. This also makes our method include rotational invariance. Since the proposed model involves both $\text{L}_{2,p}$ -norm maximization and $\text{L}_{2,s}$ -norm minimization, it is very challenging to solve. To address this problem, we present an efficient nongreedy iterative algorithm. Besides, motivated by trace ratio criterion, a mechanism of automatically balancing the contributions of different terms in our objective is found. RDS is very flexible, as it can be extended to other existing feature extraction techniques. An in-depth theoretical analysis of the algorithm’s convergence is presented in this article. Experiments are conducted on several typical databases for image classification, and the promising results indicate the effectiveness of RDS. [ABSTRACT FROM AUTHOR]