| 摘 要: 交通标志识别作为典型的机器视觉应用，已有多种机器视觉算法得到广泛的应用。卷积神经网络能够避免 显式的人工特征提取过程，因此本文引入卷积神经网络为交通标志进行识别研究，并与BP神经网络、支持向量机进行 对比实验，通过对实验结果的理解与分析，可以得出卷积神经网络在识别率及训练速度上均显著高于另两种算法，并能 取得最佳的识别效果。 |
| 关键词: BP神经网络；支持向量机；卷积神经网络；交通标志 |
|
中图分类号: TP393.0
文献标识码: A
|
| 基金项目: 基于多源探测的室内公共场所火宅检测，辽宁省科技厅（201402116）；基于子空间分析的交通标志识别研究，国家自然科学基金（61540069）. |
|
| Comparison and Analysis of Traffic Signs Recognition Algorithm |
|
ZHONG Ling，YU Yajie，ZHANG Zhijia，JIN Yongchao
|
( Shenyang University of Technology, Shenyang 110023, China)
|
| Abstract: Traffic signs recognition as a typical machine vision application,a variety of machine vision algorithms have been widely used.Convolutional neural network can avoid explicit artificial feature extraction process.Therefore,this thesis introduces convolutional neural network for traffic sign recognition research,and comparative experiments with BP neural network,support vector machine,through the understanding and analysis of the experimental results,it can be derived from the convolution neural network in recognition rate and the training speed were significantly higher than those of the other two algorithm, and can achieve the best effect of recognition. |
| Keywords: BP neural network;support vector machine;convolutional neural network;traffic signs recognition |
