6533b82ffe1ef96bd1296311

RESEARCH PRODUCT

Pulse rate variability measurement with camera-based photoplethysmography

subject

description

Electrocardiogram (ECG) has been used by doctors and biomedical researchers to measure cardiac parameters such as Heart Rate (HR) and Heart Rate Variability (HRV). HR is a medical index for health monitoring and the HRV is a sign to reflect the activities of Autonomic Nervous System (ANS) and can be used for emotion recognition applications. Recently, remote photoplethysmography (rPPG) has evolved as a non-contact technique for measuring vital cardiac signs. Compared with ECG, this technique is non-invasive, low-cost, comfortable and possibly utilized in long-term monitoring. It has great potential in remote health assessment and emotion detection. However, the rPPG is a video-based method, thus the measurement is not precise and the performance is heavily affected by the image noise, sensor noise, light variation, head movement, etc. Therefore, this method should be carefully studied and improved. In this manuscript, we have focused on two major issues for the rPPG method. Firstly, the selection of region of interest (ROI) is a critical step of the technique to obtain reliable pulse signals. It should contain as many skin pixels as possible with a minimum of non-skin pixels. Secondly, as a possible replacement of HRV in some conditions, the Pulse Rate Variability (PRV) is more complicated to measure than HR because it is then necessary to detect the peaks on the temporal rPPG signal, which is usually quite noisy and has a lower temporal resolution than the signals obtained by contact equipment. Since the PRV signal is important for various applications such as remote recognition of stress and emotion, the improvement of the PRV measurement in rPPG framework is a critical task. In the PhD thesis, we firstly introduce the scientific background for the cardiac parameter measurements and related research works. Then we describe the four contributions we have made. The first contribution is the comparative study for several ROI segmentation methods and color channel selection methods. We have identified the best combination of these methods. The second contribution is a novel method for improvement of the ROI detection method. We test the algorithm in the framework of HR measurement and show that it performs better than existing methods. The third and fourth contributions are the improvement of the remote measurement of PRV with a novel peak detection method based on one-window and two-window methods respectively.