基于改进YOLOv3和BGRU的车牌识别系统  

License plate recognition system based on improved YOLOv3 and BGRU

作　　者：史建伟 章韵[1] SHI Jian-wei;ZHANG Yun(School of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China)

机构地区：[1]南京邮电大学计算机学院,江苏南京210023

出　　处：《计算机工程与设计》2020年第8期2345-2351,共7页Computer Engineering and Design

摘　　要：针对传统中文车牌识别方法准确率不高、速度慢的问题,提出一种在自然交通场景下进行车牌定位和识别的端到端的深度学习模型。在卷积神经网络上扩展多尺度检测的深度,改进YOLOv3原有的检测网络,提升车牌对小物体的定位精度;系统利用BGRU优化识别网络,完成对定位车牌的无字符分割的识别任务,明显缩短训练时间,提升网络的收敛速度和识别准确率。实验结果表明,相比现存的传统车牌识别技术,改进方法极大地提高了车牌识别准确率和速度,鲁棒性和可靠性较好。Aiming at the problem that the traditional Chinese license plate recognition method lacking of accuracy and speed,an end-to-end deep learning model for license plate location and recognition in natural traffic scenarios was proposed,and the depth of multi-scale detection was extended on the convolutional neural network,for the purpose of improving the original detection network of YOLOv3 and positioning accuracy of the license plate for small objects.BGRU was used to optimize the recognition network to complete the character segmentation-free recognition task of the license plate,significantly shortened the training time and improved the convergence speed and recognition accuracy of the network.Experimental results indicate that compared with the existing traditional license plate recognition technology,the improved method greatly enhances the accuracy and speed of license plate recognition,and it has better robustness and reliability.

关 键 词：深度学习 车牌定位 车牌识别 端到端 无字符分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]