基于双通道动态像素聚合的交通标志识别算法  

Traffic Sign Recognition Algorithm Based on Dual-channel Dynamic Pixel Polymerization

作　　者：王杨 王傲 许佳炜 马唱 谷天祥 WANG Yang;WANG Ao;XU Jia-wei;MA Chang;GU Tian-xiang(School of Computer and Information,Anhui Normal University,Wuhu 241000,China;School of International Exchange and Cooperation,Wuhu Institute of Technology,Wuhu 241000,China)

机构地区：[1]安徽师范大学计算机与信息学院,安徽芜湖241000 [2]芜湖职业技术学院国际教育管理学院,安徽芜湖241000

出　　处：《计算机技术与发展》2023年第6期41-46,共6页Computer Technology and Development

基　　金：国家自然科学基金项目(61871412);安徽省高校自然科学研究重点项目(KJ2019A0979)。

摘　　要：交通标志自动识别有助于自动驾驶车辆自主感知外部复杂环境中的交通标识,以辅助驾驶员应对复杂路况,从而避免交通事故的发生。针对现存交通标志自动识别算法存在识别效率和准确率不高的问题,采用了一种基于双通道动态像素聚合的交通标志识别算法。首先,采用色调-饱和度-明度(Hue Saturation Value,HSV)颜色空间模型提取交通标志的颜色特征;其次,采用自适应阈值选取策略的Canny边缘检测算法分割交通标志图像的前景和背景提取交通标志的形状特征;接着,将提取到的两个物理特征进行融合,并通过反向传播(BP)神经网络进行学习训练。实验表明,该算法的交通标志识别率为95.34%、平均识别时间为1.32 ms。与已有相关算法相比,该算法不仅能够提高交通标志识别的准确率,而且在识别效率上也有一定的提高。Automatic identification of traffic signs helps self-driving vehicles to independently perceive traffic signs in external complex environments to assist drivers in coping with complex road conditions and to avoid traffic accidents.Aiming at the low recognition efficiency and accuracy of the existing traffic sign automatic recognition algorithm,a traffic sign recognition algorithm based on dual-channel dynamic pixel polymerization is adopted.Firstly,the HSV(Hue Saturation Value)color space model is used to extract the color characteristics of traffic signs.Secondly,the Canny edge detection algorithm of adaptive threshold selection strategy is used to segment the foreground and background of traffic signs images,which takes the shape characteristics of traffic signs.Then the two physically features extracted are fused to learn and train through the back propagation(BP)neural network.Experiments show that the traffic sign recognition rate of the proposed algorithm is 95.34%,and the average recognition time is 1.32 ms.Compared with the existing relevant algorithms,the proposed algorithm can not only improve the accuracy of traffic sign recognition,but also improve the recognition efficiency to a certain extent.

关 键 词：交通标志识别 特征融合 CANNY边缘检测 HSV颜色模型 BP神经网络 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]