基于扩散性搜索区域的车道线检测与跟踪算法  

An algorithm of lane detection and tracking based on diffusibility search region

作　　者：莫建文[1] 范楷[1] 张顺岚[1] 

机构地区：[1]桂林电子科技大学信息与通信学院,广西桂林541004

出　　处：《桂林电子科技大学学报》2011年第6期464-468,共5页Journal of Guilin University of Electronic Technology

基　　金：国家自然科学基金(61062003;61162007;60802018);广西科学研究与技术开发项目(桂科攻0996028;桂科攻11107006-45);广西自然科学基金(2011GXNSFA018158)

摘　　要：针对现有的车道线检测与跟踪算法在复杂环境下实时性和鲁棒性较差的问题,提出了基于扩散性搜索区域的车道线检测与跟踪算法。该算法在搜索车道线的起始阶段引入了扩散性搜索区域,搜索过程中对该区域进行不同层次的划分,缩小了算法搜索范围,提高了算法检测速度及准确率;跟踪过程中将扩散性搜索区域与动态感兴趣区域相结合,使算法在最小的区域内得到足够的车道线信息,保证了算法的实时性。对实测数据进行仿真,结果表明该算法正确检测率高、实时性好、鲁棒性强。To improve real-time and robustness in lane detection and tracking,an algorithm based on diffusibility search region was proposed.Diffusibility search region was introduced in the initial stage of lane search.The region was divided according to different purpose in the course of searching.The scope of incipient lane search was reduced and the detection speed and accuracy of the algorithm were improved.The diffusibility search region was combined with the dynamic region of interest（ROI） in stage of lane tracking.Enough information of lane was obtained in the smallest region so that the real-time of the algorithm was ensured.The simulation results for real measurements show that the proposed algorithm has high correct detection rate and good real-time and robustness.

关 键 词：车道线检测 扩散性搜索区域 感兴趣区 鲁棒性 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]