Continuous Wavelet vs. Fourier Transform-based Feature Extraction for Recognizing Sleepiness from Steering Behavior

INTRODUCTION — Developing fatigue monitoring devices, which are cheap, non-intrusive, and robust even under extreme demanding environmental conditions (e.g. high background noise, temperature, or humidity) still remains a challenging task. Utilizing steering wheel behaviour might potentially fulfil these demands [1–3]. However, little empirical research has been done to examine thoroughly the benefits of more recently feature extraction methods as e.g. the wavelet analysis [4]. Hence, the aim of this study is to apply multiple state-of-the-art pattern recognition methods [5–7] on wavelet based steering features to detect sleepiness.