Driver Fatigue Detection Algorithm Based On Driver Eye States

Fatigue driving is the most risk factor in a traffic accident and the first cause in traffic accidents death. So the research of fatigue driving test early warning system has a very important meaning at avoiding traffic accidents and enhancing traffic safety.On the basis of some research abroad, lots of methods was detected and analyzed and some real-time fatigue driving test methods based on visual type was also deeply studied. In this paper, I make a survey of a key technology based on visual fatigue driving test, including face detection,locating and tracking eyes,identification of the characteristic of eyes and analysis of mental fatigue:(1)According to the existing research face detection methods and the features of skin color space clustering, we use ycbcr color space as a color distributed statistical projections space and establish two-dimensional gauss based on the color distribution mathematical model. At the same time, we ues the method based on similar of degrees and face shape of man to detect and locate the face of the testing.(2)Face tracking. We use carl's forecast the target time in the current the initial location and then we use Mean shift in the location of domain for the next goal. At the same time we use bhattacharyya coefficient to measure the tolerance between"objective model"and"candidate model", and then ensure that whether change"candidate model"to"objective model", in order to avoid interference in the invalidation shading.(3)With the summary of existing eye positioning methods, we use the horizontal projection methods to rough locate eyes, and then we use the method based on gray- level integral projection and template matching to exactly locate eyes. And two-way integration and the law of the analysis and comparison, the ultimate test for this method is used to detect the eye of the high degree of accuracy and feasibility. Finally, we use the method of calculation perclos to test driving fatigue.The experiment finds that the method"Driver Fatigue Detection Algorithm Based on Driver Eye States"can exactly test the state of the driver and it satisfy real– time,noninvasive and high sensitivity of the systerm. This research is of profound theoretical and practical significance.