Driver Fatigue Detection Based on Real-Time Eye Gaze Pattern Analysis

This paper introduces a real time non-intrusive method to determine driver fatigue by analyzing eye gaze patterns. Using a standard webcam and a personal computer, the proposed method combines different techniques in order to keep a low computational cost without a loss of performance. Facial features are identified in a reference image to extract a region of interest, around the eyes of the user, and tracked by an optical flow algorithm in subsequent frames. Color segmentation on the resulting images allow the system to extract data needed to determine ocular following, blink detection, frequency, and percentage of eye lid closure over time (PERCLOS). This approach, while simple, proves to be very efficient and accurate for the hardware restricted setup, allowing faster information processing on modest specifications systems. For safety reasons, our experiments are limited to different subjects, simulating fatigue in laboratory conditions as well as a real time test on a moving vehicle to analyze the blinking patterns.