【摘要】 在智能交通系统中,基于视频的多车道划分是实现车流量检测、车辆跟踪及车队长度计算等的重要前提。传统方法是采用Hough变换检测车道标志线来实现车道的划分,但在复杂背景下该算法检测车道线时会产生车道线间断、干扰直线和检测不准确的现象。针对这一问题,本文提出了一种复杂背景下多车道线划分的新方法。根据传统Hough变换粗略提取出的各车道标识线附近的所有亮点像素构成特征样本集;按照模糊聚类原则划分出每条车道线所属的模糊子集;建立隶属函数确定直线在参数空间中的变量,实现车道标识线的精确检测。实验证明,本文算法能够准确地划分出各车道,且具有良好的鲁棒性。更多还原

【Abstract】 Multi-lane partition based on video is important premise of vehicle flow detection,vehicle tracing and calculation of the motorcade length in the intelligence transportation system.The traditional partition method is implemented by using Hough transform to detect lane marking line,but it is not suitable for the complex road conditions.The result of partition is affected with lane line interruption,line interference and detection inaccurateness.According to this problem,a new method of lane marking line extraction is proposed in complex environment.Hough Transform is firstly used to get initial lane marking line.Then,Fuzzy set is divided according to fuzzy clustering principle.At last,Membership function is established to get variable in the parameter space to realize precise detection of lane marking line.It is proved that this algorithm can divide lanes accurately and has the good robustness.更多还原