| 基于卷积神经网络LeNet-5的车牌字符识别研究 |
| |
| 引用本文： | 赵志宏,杨绍普,马增强.基于卷积神经网络LeNet-5的车牌字符识别研究[J].系统仿真学报,2010,22(3). |
| |
| 作者姓名： | 赵志宏 杨绍普 马增强 |
| |
| 作者单位： | 1. 北京交通大学机械与电子控制工程学院,北京,100044;石家庄铁道学院,河北,050043
2. 石家庄铁道学院,河北,050043 |
| |
| 基金项目： | 国家杰出青年科学基金(50625518); 教育部科学技术研究重点项目(205019) |
| |
| 摘 要： | 将卷积神经网络LeNet-5引入到车牌字符识别中。为了适应目前中国车牌字符识别的需要,对传统的卷积神经网络LeNet-5的结构进行了改进,主要是改变输出单元的个数与增加卷积层C5特征图的个数。研究结果表明,改进后的LeNet-5比传统的LeNet-5的识别率有所提高,识别率达到98.68%。另外,与BP神经网络进行了比较研究,从实验中可以看出在字符识别的正确率和识别速度上都优于BP神经网络。卷积神经网络在车牌识别中具有很好地应用前景。
|
| 关 键 词： | 字符识别 车牌识别 卷积神经网络 LeNet-5 |
License Plate Character Recognition Based on Convolutional Neural Network LeNet-5 |
| |
| ZHAO Zhi-hong,YANG Shao-pu,MA Zeng-qiang.License Plate Character Recognition Based on Convolutional Neural Network LeNet-5[J].Journal of System Simulation,2010,22(3). |
| |
| Authors: | ZHAO Zhi-hong YANG Shao-pu MA Zeng-qiang |
| |
| Institution: | ZHAO Zhi-hong1,2,YANG Shao-pu2,MA Zeng-qiang1,2 (1. School of Mechanical,Electronic , Control Engineering,Beijing Jiaotong University,Beijing 100044,China,2. Shijiazhuang Railway Institute,Shijiazhuang 050043,China) |
| |
| Abstract: | The application of convolutional neural network LeNet-5 was proposed in license plate character recognition. To fit with the Chinese license plate character recognition problem, the traditional LeNet-5 was modified. The unit number of output layer was changed and the feature map number of C5 layer was added. Experimental results show that the recognition rate of modified LeNet-5 reaches 98.68% and is better than that of LeNet-5. The results are also compared with the BP neural network, which indicates that ... |
| |
| Keywords: | character recognition license plate recognition convolutional neural network LeNet-5 |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
