Driver Fatigue Warning Research Based On Eye Detection

In recent decades, with the rapid development of technology and productivity levels, vehicles rose around the world every year. A variety of means of transport not only bring great convenience to humans but also produced a huge number of traffic accidents, caused incalculable loss to human lives and property. Traffic safety is a growing concern of the whole society. Surveys have shown that in many factors that lead to accidents, the driver fatigue accounts for a sizable proportion. So, a real-time, accurate detection of driver fatigue warning system driver fatigue was developed urgently.On the basis of a large number of domestic and foreign fatigue driving warning research, this paper choose the study of fatigue driving warning based on machine vision with strong real-time, accuracy and non-contact advantages. The main objective of this article is to achieve the realization of driver fatigue driving warningAccording to the processing result on the image information (mainly eyes state information of driver fatigue detection mechanism) captured by the on-board video camera combine with proper fatigue judgment mechanism.. This article describes some important techniques in fatigue detection in details including image pre-processing, face detection and location, the human eye detection and location, the state of the human eye fatigue extraction and determination.The main work of this paper is as follows:1. Details of the various digital image pre-processing methods, including image graying, histogram equalization, de-noising, etc., and focuses on the image edge detection and binarization principles and algorithms. Image pre-processing plays an important role in digital image processing, for the below face and eye detection and location basis.2. After comparing the two commonly used face detection method:face detection based on facial features and face detection based on the model. This article selected color gray projection method combined paper-based methods as face detection algorithms. At the same time focuses on the use of combined YCbCr color space and HSV color space color modeling method for face skin color region segmentation, and then remove the neck and other locations gray integral projection method based on skin color region, and ultimately mark a person’s face and realize people face detection and location.3. According to idea of the human eye detection through rough location to pinpoint, proposed a combination of gray Integral Projection and Hough transform algorithm for detection and location of the human eye. Start with a two-way split of the gray integral projection box set eyes and the eyes of the region outline, and then by Hough transform in the eyes contour detection pupil of the human eye to achieve precise positioning. Experimental results show that this method can achieve basic human eye detection and location, but it also has some drawbacks to be optimized and improved.4 Highlights PERCLOS principle, determine the values and methods combined PERCLOS blinking frequency of the human eye to locate the determination of fatigue and fatigue judgment based on an analysis of the statistical results of the experiment, and ultimately determine fatigue.Huang Xiong (Control Engineering) Supervised by Professor Yang Huicheng and Senior Engineer Fang Yunzhou.