摘要
本文对基于机器视觉的路径识别方法进行了研究。通过对采集的路面图像进行中值滤波、边缘增强、动态阈值二值化,获得良好的路标图像。根据道路特征识别出道路标识线,并拟合出车辆运动的引导线。使用梯形小视野直线拟合法,减少图像处理时间和提高道路识别的可靠性。通过自制小车的实验表明,该算法具有良好的实时性、可靠性。
This paper presents a fast vision oriented algorithm for automatic lane detection. In order to get good lane maker images, median filter, edge enhancement and automatic threshold are adapted to processing the images taken by CCD camera. Lane marker is identified according to its feature, and directional line for vehicle motion is generated based on the lane maker. By using little trapeziform AOI, the image processing time is reduced and the reliability of lane detection is improved. Experiment proves the high efficiency and reliability of the new method.
出处
《微计算机信息》
北大核心
2006年第09Z期197-200,共4页
Control & Automation
基金
总装备部预研项目(40404070302)
关键词
机器视觉
路标检测
梯形视野
引导线
machine vision,lane marker detection,trapeziform AOI,directional line
