车道线识别中的自适应Canny边缘检测及改进研究  

Achievement and Improvement of Adaptive Canny Operator in Lane Identification

作　　者：陈阳舟[1] 闫豪杰[1] 辛乐[1] 

机构地区：[1]北京工业大学自主技术与智能控制研究中心,北京100124

出　　处：《交通信息与安全》2012年第1期148-151,共4页Journal of Transport Information and Safety

摘　　要：针对Canny边缘检测阈值在车道线识别中不易选取的问题,提出了基于Otsu算法实现自适应Canny边缘检测的方法。实验验证表明,其对远视野道路图像可以获得良好的边缘检测效果,而对近视野道路图像效果较差。进一步提出了目标区域补偿策略改进上述算法。结合Hough变换算法,实现了车道线的识别。实验结果表明,改进的算法可以达到实时获取车道线的要求。For the difficulty of selecting the segmentation threshold in the process of Canny operator-based lane identification,a method of adaptive segmentation threshold in virtue of Otsu algorithm is proposed.Experiments to verify the effectiveness of the algorithm show that it can get good binarization results for far field images,but unavailability for close vision images.Therefore,further improvement is made by proposing a target area compensation strategy.Finally,combining with Hough transform,the lane recognition is ultimately achieved.The Experimental results show that the improved algorithm meet real-time performance of lane detection and identification.

关 键 词：车道线识别 CANNY边缘检测 OTSU算法 目标区域补偿 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]