Computer Engineering and Applications ›› 2019, Vol. 55 ›› Issue (18): 195-200.DOI: 10.3778/j.issn.1002-8331.1806-0155

Previous Articles     Next Articles

Traffic Sign Recognition Algorithm Based on Multiple-Feature Fusion

HAN Xixi, WEI Min, XU Xiyi, LI Qiaoyue, CHEN Xi, ZHU Hancheng   

  1. 1.School of Information and Control Engineering, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China
    2.Communication Information Center, Shandong Energy Xinwen Mining Group, Tai’an, Shandong  271213, China
    3.Zhaizhen Coal Mine, Shandong Xinwen Mining Group, Xintai, Shandong  271204, China
  • Online:2019-09-15 Published:2019-09-11

基于多特征融合的交通标志识别算法

韩习习,魏民,徐西义,李巧月,陈曦,祝汉城   

  1. 1.中国矿业大学 信息与控制工程学院,江苏 徐州 221116
    2.山东能源新汶矿业集团 通信信息中心,山东 泰安 271213
    3.山东省新汶矿业集团 翟镇煤矿,山东 新泰 271204

Abstract: Taking recognition rate, time complexity and robustness all into consideration, this paper proposes a traffic sign recognition algorithm based on edge, texture and color feature fusion and Support Vector Machine(SVM). It conducts the statistical average on extracted Histogram of Oriented Gradient(HOG) features which can describe the edge information of traffic signs images. The reduced-dimension HOG-maxLBP features are obtained by fusing with Local Binary Pattern(LBP) features that can represent traffic sign internal texture information. The color features are concatenated with HOG-maxLBP features as the final features. Traffic sign training and classification are performed using SVM. Experimental results show that the proposed algorithm not only improves the traffic sign recognition rate, but also reduces the time complexity and enhances the system robustness.

Key words: Traffic Sign Recognition(TSR), Histogram of Oriented Gradient(HOG), Local Binary Pattern(LBP), color features, feature fusion, Support Vector Machine(SVM)

摘要: 综合考虑识别率、时间复杂度以及鲁棒性,提出一种边缘、纹理、颜色多特征融合和支持向量机(SVM)的交通标志识别算法。通过提取能够描述交通标志图像边缘信息的方向梯度直方图(HOG)特征并进行统计平均,与能够表示标志图像内部纹理信息的局部二值模式(LBP)特征融合得到降维后的HOG-maxLBP特征,再级联交通标志的颜色特征作为最终的特征向量,最后利用SVM进行交通标志训练和分类。实验结果表明,该算法不仅提高了交通标志的识别率,而且降低了时间复杂度,增强了系统鲁棒性。

关键词: 交通标志识别(TSR), 方向梯度直方图(HOG), 局部二值模式(LBP), 颜色特征, 特征融合, 支持向量机(SVM)