Preliminary development and test of a new automatic drowsiness quantification system using range and intensity images obtained from a dashboard-mounted near-infrared 3D range sensor

Drowsiness is the cause of a variety of accidents, in particular in transportation. It is thus critical to develop systems that can monitor the level of drowsiness of a vehicle operator, such as a driver, automatically and continuously so as to take timely safety measures. One of the least intrusive approaches for this is to use one or more cameras mounted in the vehicle dashboard. We report here on the preliminary development and test of such a system that uses ocular parameters extracted from the images obtained from a dashboard-mounted camera consisting of a 3D range imager, namely the Kinect v2 sensor, which provides near infrared (NIR) intensity images and range images. We use the range images to construct 3D deformable models of the face, and the NIR intensity images to track the face and to analyse the facial expressions, including in darkness.