基于方向边缘匹配的人行横道与停止线检测  

Pedestrian Crossing and Stop Line Detection Based on Oriented-edge Match

作　　者：谢锦[1,2] 蔡自兴[1] 李昭[1] 尤作[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083 [2]湖南师范大学工学院,长沙410081

出　　处：《计算机工程》2013年第6期261-265,282,共6页Computer Engineering

基　　金：国家自然科学基金资助重大项目(90820302);国家自然科学基金资助项目(60805027);博士点基金资助项目(200805330005)

摘　　要：针对智能车辆视觉导航系统的路面标线实时感知问题,提出基于方向边缘匹配实现人行横道和停止线的检测算法。利用逆透视映射表快速建立车辆前方感兴趣区域俯视图,提取图像垂直边缘和水平边缘,通过亮度上升与下降相邻边缘匹配检测定位人行横道与停止线。实验结果表明,该检测算法满足实时性要求,具有强鲁棒性,能应用于复杂城市道路环境。In order to solve the real-time road marking recognition problem in the intelligent vehicle vision navigation system, a novel pedestrian crossing and stop line detection algorithm based on oriented-edge match is proposed. It builds top-view image through Inverse Perspective Mapping（IPM） transformation by a mapping table to accelerate IPM process. It abstracts vertical oriented-edge and horizontal oriented-edge. The pedestrian crossing and stop line can be detected by matching adjacent up and down edges to bands. Experimental results illustrate that the algorithm is effective for real-time analysis and robust under difficult urban road circumstances.

关 键 词：方向边缘 逆透视映射 感兴趣区域 线性拟合 匹配 随机抽样一致性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]