基于特征颜色的车道线检测算法  

Lane Detection Algorithm Based on Feature Color

作　　者：胡骁[1] 李岁劳[1] 吴剑[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072

出　　处：《计算机仿真》2011年第10期344-348,共5页Computer Simulation

摘　　要：研究自动车驾驶车道优化识别特征颜色边缘,针对不同颜色车道线检测,精度算法不完善。为了提高准确性,提出一种基于特征颜色边缘检测算法。根据道路图像不同物体颜色特征,按特征颜色设计转移函数来标记原图像中的特征颜色区域,再对选取的特征颜色区域进行边缘检测并输出二值化,进行识别过程,对结果进行边缘细化操作得到最终的检测结果。和其它特征颜色的边缘检测算法对比表明,改进的算法在边缘识别上,提高了定位的准确度和精细度,实时性地满足自动驾驶系统的要求,对自动车辆车道线检测有实际参考意义。The paper proposed a lane detection algorithm based on feature color.By analyzing the color characters of different objects in lane images,we designed a transform function based on the feature color to mark the specific color region in source images,then extracted the edges in the selected region and outputted the binary results.To improve the accuracy and get thinner edges,an edge thinning technology was applied.Compared with other algorithms of feature color edge detection,the proposed algorithm is more accurate.Time consumption can satisfy the need of automatic driving system,so it has realistic significance for lane detection in the automatic driving system.

关 键 词：自动驾驶 特征颜色边缘 车道线检测 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]