| With the popularization of automobiles and highways, the performance of the automotive active safety has been also improved, and the driveway deviates of the early-warning system are just one of the main security products. Based on the machine vision, the road identification and obstacle detection are taken as the research objects in this thesis. By means of the dynamic image process, the state information about the environment and the traffic condition is provided for driving vehicles. Through comprehensive utilization of the current driveway state (curved way or straight way) and the distance information, the visual navigation aim for vehicles is realized.In this thesis, the driveway deviates navigation system based on the vision is discussed in detail. This system can be divided into three basic modules: the module of the road and vehicle state perception, the driveway detection algorithm and the interface of the signal demonstration. The module of the road and vehicle state perception, and the driveway detection algorithm are taken together as video recognition system, which is the main part of the navigation system for deviated automobiles. The main results in this thesis are as follows:1. The video recognition system is established according to following steps:(1) to preprocess images(2) to study the detection algorithm for the driveway boundary(3) to dismember the images by means of the optimal threshold value(4) to recognize the edges or the divided driveways by means the detection of the divided driveways(5) to establish the driveway model2. Based on the combination of the video recognition system and the visual interfaces, the vehicle navigation system has been established. |
