卷积神经网络在车牌识别中的应用实现  

Implementation of convolutional neural networks for licence plate recognition application

作　　者：陈明洋 代红[1] 王思彤 陈禹含 CHEN Mingyang;DAI Hong;WANG Sitong;CHEN Yuhan(University of Science and Technology Liaoning,Anshan 114000,China)

机构地区：[1]辽宁科技大学,辽宁鞍山114000

出　　处：《无线互联科技》2024年第22期62-64,共3页Wireless Internet Science and Technology

基　　金：辽宁科技大学省级大创训练项目:项目名称:停车场智能分析与预测。

摘　　要：随着交通管理、智慧城市等领域的快速发展,车牌识别技术逐渐成为一种关键技术。卷积神经网络作为一种强大的图像处理和特征提取方法,被广泛应用于车牌识别领域。文章采用机器学习、字符分割等技术以提高车牌识别的高效性和准确性。系统涵盖了多种情况下的车牌图像并对图片进行精密的预处理。在车牌分类识别方面,文章除了运用字符分割与关键字识别技术,还设计了具有针对性的卷积神经网络模型,对其进行训练并在多组对比实验后,对模型的识别准确率进行评估。结果表明:卷积神经网络具有较好的识别准确率和一定的实用性和价值性,可应用于智能停车场的车辆识别。With the rapid development of traffic management,smart city and other fields,the license plate recognition technology has gradually become a key technology.The convolutional neural network,as a powerful image processing and feature extraction method,is widely used in the field of license plate recognition.Meanwhile,this study also uses machine learning,character segmentation and other techniques to improve the efficiency and accuracy of licence plate recognition.The system covers license plate images in various situations and performs sophisticated preprocessing of the images.For license plate classification and recognition,besides using character segmentation and keyword recognition techniques,a targeted convolutional neural network model is designed and trained.The recognition accuracy of model is evaluated after the multiple sets of comparison experiments.The convolutional neural network is applied to vehicle recognition in intelligent parking lots in this study,which has good recognition accuracy and certain practicality and value.

关 键 词：卷积神经网络 车牌识别 机器视觉 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程]