Main Article Content
Abstract
A Driver Pattern Recognition System was developed, using concepts based on non-intrusive machine vision. The device uses a small monochrome security camera which points directly to the driver's face and monitors the driver's eyes for fatigue detection. In such a case when fatigue is detected, the driver is alerted by a warning signal as well as if the driver is distracted it will also alert the driver to be more careful. This report describes how eyes can be found, and also how to determine whether eyes are open or closed. The developed algorithm is unique to any currently published papers which were the project's primary objective. The device deals with finding the edges of the face using information obtained for the binary version of the image, which narrows the area of where the eyes will exist. When the face area is defined, the eyes are located by measuring the area's horizontal averages. Bearing in mind the knowledge that eye regions on the face present major changes in intensity, The eyes are found by detecting major changes in the facial pressure. Once the eyes are positioned, measuring the distances between the changes in intensity in the area of the eye determines whether the eyes are open or closed. A great distance corresponds to the closure of the eyes. If the eyes are located closed for five consecutive frames, the machine assumes that the driver is falling asleep and sends out an alarm signal. Also, the system can detect when the eyes can not be found and work under reasonable conditions of light.