A Study Of Eye's Location Approach On Driver Fatigue Detection

With the development of society and economy, more and more mobile vhiecles in our society, at the same time, traffic accidents have become the current serious problems for all countries, and fatigue driving is one of the prime factors that caused a traffic accident. Compared with other methods of Fatigue detection, using Machine Vision has great advantages in real time, non-contact and all-weather monitoring, so this method has become a hotspot of current research.On the basis of previous work, Aim at fatigue detection of driver, this paper main study the method of eyes'location and pupils'location. It consists of four processes:human face detection and location, human eyes'location, pupils'detection and location of human eyes and identify the state of pilot fatigue. In this paper, the content and main results are as follows:(1) Human face detection and location. Grab an infrared image from the infrared camera of the video stream, and then make the use of the advantages of the higher brightness of infrared human face region, dark background and easy, and binarize the image with the improved iterative thresholding algorithm, and then locate the face region with the method of region marked based on exploring and searching. The face detection method,which is relative simple, can not only locate the face accurately, but also basically free from the effects of light. It resolves the disadvantages of the tradition detection which is unable to locate the face accurately owing to some obstacles, such as unstable light,long hairs,long fringe etc.(2) Human eyes'location. This algorithm is based on getting human face, and making full use of Human eyes'complexity and geometry feature of three chambers and five holes; and then get the smallest eyes-block by RAMF(Ranked-order Adaptive Median Filtering) method and iterative threshold choosing algorithm. Finally, getting the eyes' position accurately according to the centroid of the eyes-block and integral projection function.It's effective to avoid some obstacles,such as long fringe, long hair and so on.(3) Pupils'detection and location.This paper carefully analyze the feature of Bright Pupil and SUSAN(Smallest Univalue Segment Assimilating Nucleus) model,and propose the Circle-corner point;and then use the improved SUSAN algorithm to find circle-corner point, computing the gradient value of candidate pupil points according to magnifing and narrowing of gradient; finally, output the center of pupil and the area of Bright pupil by selection function. According to the PERCLOSE principle, combining pupil location with the output of pupil's area, this paper adopt the calculation of pupils'area to detect the state of drivers'fatigue in real time. So this method make full use of the pupils'feature and has the advantages with accurate location, high speed, and can satisfies the request of real time.