基于计算机视觉高速智能车辆的道路识别

Lane Detection for High Speed Intelligent Vehicle Based on Machine Vision

论文研究了基于计算机视觉高速智能车辆的道路识别。通过对JLUIV-4智能高速车辆系统采集的图像进行中值滤波、边缘增强、最优阈值二值化,获得良好的梯度图像。根据道路特征采用Hough变换识别出道路边界。使用感兴趣区域,减少图像处理时间和提高道路识别的可靠性。JLUIV-4的高速导航实验表明,该算法具有良好的实时性、可靠性和鲁棒性。

The paper presents a fast and rob us t approach of automatic lane detection for high speed intelligent vehicle.In o rder to obtain good gradient images,median filter,edge enhancement and optim al threshold are adopted to process-ing images taken by JLUIV-4CCD camera.La ne edge is located according to it's feature model through Hough trans-formatio n.By focusing on Area of Interest(AOI )processing time is dramatically redu ced.Moreover,the reliability of lane detection is improved.The experimental r esult shows that this approach is of high efficiency,reliability and robustnes s on JLUIV-4.