基于深度学习的汽车车牌识别算法研究  

Research on Vehicle License Plate Recognition Algorithm Based on Deep Learning

作　　者：高昕葳 Gao Xinwei(Department of Mechanical Electronical and Engineering,Gansu Forestry Technological College,Tianshui,Gansu 741020,China)

机构地区：[1]甘肃林业职业技术学院机电工程学院,甘肃天水741020

出　　处：《机电工程技术》2021年第10期164-166,共3页Mechanical & Electrical Engineering Technology

摘　　要：随着我国汽车的保有量逐渐增加,车牌识别在智慧车辆管理系统中起着重要作用。现有的车号识别算法识别速度慢、准确度不高,容易受光线及车牌位置角度与摄像机相对固定位置的影响而造成误识别。基于深度学习的Faster-RCNN进行车牌定位,生成车牌提取框提取车牌;使用VGG16网络模型识别字符,最终完成汽车车牌的识别。在大量的数据集中进行训练、测试,仿真结果表明在复杂环境下采用Faster-RCNN与VGG16结合的网络模型对车牌的识别准确率高达99.2%,识别准确率优于其他算法。With the gradual increase in the number of cars in my country,license plate recognition plays an important role in the smart vehicle management system.Existing vehicle number recognition algorithms have slow recognition speed and low accuracy,and are susceptible to misrecognition due to the influence of light,the angle of the license plate position and the relative fixed position of the camera.Firstly,the license plate was located based on the Faster-RCNN of deep learning,the license plate extraction frame was generated to extract the license plate.Secondly,the VGG16 network model was used to recognize characters and finally complete the recognition of the car license plate.Training and testing were carried out in a large number of data sets.The simulation results show that the recognition accuracy of the license plate using the network model combined with Faster-RCNN and VGG16 in a complex environment is as high as 99.2%,and the recognition accuracy is better than other algorithms.

关 键 词：车牌检测 深度学习 Faster-RCNN VGG16 字符识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]