Based On Monocular Vision Image Processing Research In The Auxiliary Driving System

Traffic accident casualties in China ranks first in the world for the past ten years, traffic accidents have become the first public hazard in China. In order to reduce traffic accidents, the development of vehicle safety driving assist system based on the active safety has became one of the focus of the intelligent vehicle systems. With the development of computer vision technology, active safety driver assistance system based on monocular vision has been an important research direction.This paper focuses on driving assist system based on Monocular Vision, including lane detection and tracking, vehicle detection and tracking, lane departure warning and vehicle forward collision warning. This thesis’s research is mainly in practice. This thesis’s mainly research works are as follows:1. According to the characteristics of the traffic information, split the lane detection area and the vehicle detection area, and separately using image preprocessing operation, to reduce the feature computation, to improve the detection accuracy.2. According to the structure characteristics of the highway, a lane line detection method based on the linear model was proposed, through improving the Hough transform to detect the position of the lane line. According to the correlation between adjacent video frames, proposed a method based on dynamic ROI (region of interest) to track the lane line.3. The study used a vehicle detection method based on Haar-like rectangle features and AdaBoost algorithm which are based on the correlation of classifier. Through the design of a screening filter cascade classifier can significantly improve the detection speed. According to the correlation between adjacent video frames, proposed a method based on dynamic ROI to track vehicles.4. Research principles and methods of camera calibration, the calibration of the camera’s inside parameters and outside parameters to achieve the vision measurement. Use the detected lane parameters and vehicle location information to calculate the vehicle yaw angle, the distanceb between vehicle and the left and right lane line, the distance and angle between the vehicle and the vehicle in front, proposed a lane departure warning and vehicle forward collision warning decision-making strategy.