上海交通大学学报

上海交通大学学报（自然版） ›› 2012, Vol. 46 ›› Issue (09): 1355-1360.

• 自动化技术、计算机技术 • 上一篇    下一篇

基于级联滤波的交通信号灯识别方法

金涛a，王春香a，王冰b，杨明b   

  1. (上海交通大学 a． 机器人研究所； b． 自动化系， 系统控制与信息处理教育部重点实验室， 上海 200240)
  • 收稿日期:2011-11-10 出版日期:2012-09-28 发布日期:2012-09-28
  • 基金资助:

    国家自然科学基金项目（51178268），国家自然科学基金重大研究培育计划项目（91120018），国家高技术研究发展计划（863）项目(2011AA040901)资助

Traffic Lights Recognition Based on Concatenated Filtering Method

 JIN  Tao-a, WANG  Chun-Xiang-a, WANG  Bing-b, YANG  Ming-b   

  1. (a. Research Institute of Robotics; b. Department of Automation,  Key Laboratory of System Control and Information Processing, Ministry of Education,  Shanghai Jiaotong University, Shanghai 200240, China)
  • Received:2011-11-10 Online:2012-09-28 Published:2012-09-28

摘要： 提出了基于智能车辆中车载摄像头的交通信号灯检测与识别方法.通过对已有交通信号灯图像进行训练及采用色彩分割的方法而提取候选区域；将候选区域作为输入，提出了基于级联滤波的候选区域分类方法；同时，采用标准互相关模板匹配法对级联滤波后的候选区域交通信号灯进行验证.真实环境下的实验结果表明，所提出方法在复杂的城市环境中对于智能车辆的交通信号灯识别的有效性和实时性较高.    

关键词: 交通灯识别, 色彩分割, 级联滤波, 标准互相关

Abstract: A method for detection and recognition of traffic lights based on intelligent vehiclemounted camera was proposed. Applying the threshold acquired by image training, the candidate regions of traffic lights are extracted using the color segmentation method. Next the concatenated filters are proposed as a way to classify the extracted candidate regions. And then template matching using normalized cross correlation techniques is adopted to validate the classified traffic lights candidates. The experimental results show that the proposed algorithm works effectively and robustly for traffic lights recognition for intelligent vehicles in complex urban environments.  

Key words: traffic lights recognition, color segmentation, concatenated filtering, normalized cross correlation

中图分类号: