限定交通场景中车辆行为认知方法研究

发布时间：2018-06-19 22:54

本文选题：车辆跟踪 + 车道线识别　；参考：《中国民航大学》2015年硕士论文

【摘要】：交通事故频发以及人们对交通方式越来越高的要求使智能交通系统得到了迅速发展,其中智能车辆成为最大的研究热点之一。前方运动车辆的行为认知可以为智能车辆提供实时的道路车辆信息,能够有效地避免交通意外情况的发生,已然成为无人车驾驶技术、辅助安全驾驶等领域的研究重点。为了实现准确地车辆跟踪与车道线识别,本文对一些常用的车辆行为智能识别算法进行了改进,主要贡献如下:1.提出了一种改进的前方目标跟踪方法。改进的方法充分利用了车辆尾部的对称性,有效地去除了交通环境中的干扰信息,并利用金字塔光流改善了对快速目标的跟踪效果。2.利用了一种改进Hough变换的方法实现车道线的检测。在本方法中,通过约束参数来建立动态感兴趣检测区域,在当前帧的基础上对下一帧进行参数约束Hough变换,提高车道线检测算法的准确性和实时性。3.为了便于分析车辆行为,对复杂的交通场景作了限定和建模。根据道路交通标线和车辆与车道线的位置关系,对限定交通场景进行了建模,实现了对前行车辆的保持车道、压线行驶以及车道变更等行为的认知。大量实验结果表明,文中提出的方法能够有效地在限定交通场景中实现对前方车辆的行为识别。
[Abstract]:The frequent occurrence of traffic accidents and the increasing demand for traffic mode make the intelligent transportation system develop rapidly, in which intelligent vehicle becomes one of the biggest research hotspots. The behavioral cognition of moving vehicles in front can provide real-time road vehicle information for intelligent vehicles and can effectively avoid traffic accidents. It has become the focus of research in the field of driverless driving technology and auxiliary safe driving. In order to achieve accurate vehicle tracking and lane line recognition, some commonly used intelligent recognition algorithms for vehicle behavior are improved in this paper. The main contributions are as follows: 1. An improved forward target tracking method is proposed. The improved method makes full use of the symmetry of the vehicle tail, effectively removes the interference information in the traffic environment, and improves the tracking effect of the fast target by using the pyramid optical flow. An improved Hough transform is used to detect lane lines. In this method, the dynamic detection region of interest is established by constraining parameters, and the next frame is transformed with parameter constraint Hough on the basis of the current frame to improve the accuracy and real-time performance of lane detection algorithm. In order to analyze vehicle behavior, the complex traffic scene is defined and modeled. According to the traffic mark and the position relation between the vehicle and the lane line, the limited traffic scene is modeled, and the cognition of the behavior of the moving vehicle such as keeping the lane, driving under the pressure line and changing the lane is realized. A large number of experimental results show that the proposed method can effectively recognize the behavior of the vehicle in the restricted traffic scene.
【学位授予单位】：中国民航大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP391.41

【相似文献】