Based On The Driver's Eye Tracking Monitoring Of Safe Driving

With the popularity of the car, the driver's line of sight to track is of great significance to improve the security of driving. Eye gaze tracking is also a current area of research, much attention has been paid to its in human-computer interaction, military, marketing, industrial control and other fields has a broad prospect of use. For Real-time, stability and accuracy, the author of this paper puts forward a kind of non invasive eye tracking framework. Use of a common single camera to obtain a driver's image, using the driver of the inner eye point as a reference point, iris center as a fixed point, through the position relationship between the two features eye tracking model is set up.This system the first step is filtering and contrast enhancement processing to camera image, put forward the median filter, and the combination of bilateral filtering method, and using a form of dimensionality reduction for weight calculation. In order to reduce the search space and improve the accuracy of feature extraction, the system adopts from coarse to fine search strategy. From the image after pretreatment to determine the face region, from the face region to determine the eye area, finally obtain iris center and corner features. For face recognition module, this paper puts forward a kind of based on prior knowledge combined with statistical method, using skin color model to determine the candidate face area first, then use Adaboost algorithm to check the candidate regions. This method not only shorten the checking time and improve the detection accuracy. After obtaining face region using "three court five eyes" and based on the grey value of the projection method to locate the human eye. For the center of the iris detection is proposed in this paper an improved method based on midperpendicular Hough transform, the problem of three dimensional space is transformed into a two-dimensional space. In the corner detection module, and put forward using the quadratic recursive algorithm based on harris operator, for a more precise and more detailed eye corner point detection.Finally, this article is based on openCV secondary development, testing the framework and the related algorithm. Results show that the framework of this system and the related algorithm is effective.