A monocular-vision-based driver assistance system for collision avoidance

In this paper, we describe the algorithms that how to track the feature points and how to establish the math model of lanes. An algorithm based on extend Kalman filter for dynamic target tracking is presented in this paper. Analyze the usual calculation method of distance by multi-eye, and present a real-time and efficient calculation method that makes use of computer vision theory to realize the measurement of the distance of an obstacle in the high way by single-eye. As the experiment show, the system has a good real-time property, high accuracy so as to measure up to the practical requirement This thesis studies the driving-habits discovery model of steering vehicle for a vehicle-safety expert system.