基于极性检测和改进CNN框架的车牌识别方法  

Research of license plate recognition method based on polarity detection and improved CNN framework

作　　者：李娅 王兴路 雷蕾 上官望义 Li Ya;Wang Xinglu;Lei Lei;Shangguan Wangyi(School of Vehicle Engineering,Xi'an Aeronautical University,Xi’an 710077,China)

机构地区：[1]西安航空学院车辆工程学院,西安710077

出　　处：《电子测量技术》2021年第14期26-32,共7页Electronic Measurement Technology

基　　金：陕西省教育厅专项科研计划项目(20JK0695)资助。

摘　　要：为了提高车牌自动识别的性能和适用性,提出一种基于前景极性检测和改进卷积神经网络(CNN)的车牌识别方法。所提方法包含两个主要模块:字符分割模块和字符识别模块。在字符分割模块中,基于RGB颜色的前景极性检测进行二值化处理和感兴趣区域(ROI)的分割,然后进行字符高度估计和倾斜校正等处理。在字符识别模块中,通过包含聚合模块的多通道深度CNN框架提取深度特征,提高输出特征的表征能力。实验结果表明,所提方法具有较好的识别精度,在较难的SSIG测试集和AOLP数据集上分别获得92.2%和94.1%的识别率,且在一些极端情况下优于商业的车辆门禁系统。To improve the performance and applicability of license plate recognition,a license plate recognition method based on foreground polarity detection and improved convolutional neural network(CNN)is proposed.The proposed method consists of two main modules:character segmentation module and character recognition module.In the character segmentation module,foreground polarity detection based on RGB color is used for binarization and ROI segmentation,and then character height estimation and skew correction are performed.In the character recognition module,the depth features are extracted by the multi-channel deep CNN framework including the aggregation module to improve the representation ability of the output features.The experimental results show that the proposed method has good recognition accuracy,and the recognition rate is 92.2%and 94.1%respectively on the more difficult SSIG test set and aolp data set,and it is superior to the commercial vehicle access control system in some extreme cases.

关 键 词：车牌自动识别 卷积神经网络 字符分割 字符识别 感兴趣区域 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]