作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (15): 158-160. doi: 10.3969/j.issn.1000-3428.2011.15.050

• 人工智能及识别技术 • 上一篇    下一篇

基于链码趋势表的交通灯形状识别方法

何海涛,宋 健   

  1. (四川理工学院计算机学院,四川 自贡 643000)
  • 收稿日期:2011-03-11 出版日期:2011-08-05 发布日期:2011-08-05
  • 作者简介:何海涛(1975-),男,讲师、硕士,主研方向:人工智能,嵌入式系统,图形图像处理;宋 健,讲师、硕士
  • 基金资助:
    人工智能四川省重点实验室基金资助项目“智能交通系统的图像信息处理方法研究”(2008RK011)

Traffic Light Shape Recognition Method Based on Chain Code Trend Table

HE Hai-tao, SONG Jian   

  1. (School of Computer Science, Sichuan University of Science and Engineering, Zigong 643000, China)
  • Received:2011-03-11 Online:2011-08-05 Published:2011-08-05

摘要: 设计一种交通灯箭头形状识别方法,提出链码趋势表的概念。根据交通灯颜色特征,在RGB颜色空间内进行颜色过滤和颜色加强,得到交通灯候选区域;利用轮廓跟踪算法分割图像,得到目标轮廓链码;计算链码在链码趋势表中的匹配率,结合箭头的对称特征和边长度比例特征得到交通灯箭头匹配率,并根据实验数据确定识别区间。实验结果表明,该方法的识别准确率较高。

关键词: 智能交通系统, 链码, 趋势表, 交通灯, 形状识别

Abstract: An algorithm of detecting the shape of a traffic light in a scene image and the concept of trend table is proposed. It filters and enhances color information in RGB color space by the color feather of traffic light, and gets the candidate areas, segments image using contour tracking algorithm and figures out the chain code. It calculates the matching probability by combining the symmetrical feather and length ratio feather. The discernible scale is determined by the statistics. Experimental results using videos verify the effectiveness of the proposed algorithm.

Key words: Intelligent Transport System(ITS), chain code, trend table, traffic light, shape recognition

中图分类号: