Correlation Analysis between Electrocardiography (ECG) and Photoplethysmogram (PPG) Data for Driver's Drowsiness Detection Using Noise Replacement Method

The aim of this study is to investigate a noise handling method that provides high correlation between Electrocardiography(ECG) and Photoplethysmogram(PPG) data. This issue is important in detecting driver's drowsiness. So far, there have been many studies to estimate the driver's drowsiness based on heart rate variability (HRV), by examining such features such as power spectral density (PSD) from ECG data. However, since the ECG data is obtained from heart's electrical signals through the skin's electrodes, ECG measuring instruments are inconvenient to wear in real-life driver situations. On the other hand, with the development of PPG sensors, it becomes now easy to get HRV data through smart bands which are more convenient to wear in driving situations. But the PPG data from smart bands tend to have more noise than ECG data. Thus, handling such noises is of great significance to adopt existing ECG-based methods for PPG data in driver's fatigue estimation. In this study, we propose a noise replacement method that identifies noises and substitutes them with appropriate values, not filtering out noises. From experiments, we observed that our noise replacement method enables us to obtain the improved correlation in PSD between ECG data and PPG data, compared to noise filtering method. This result implies that the noise replacement method may be a more effective way to handle PPG data for driver safety monitoring. (C) 2017 The Authors. Published by Elsevier B.V.