铁路货车车号图像识别系统的算法研究  

Algorithm Research on Image Recognition System of Railway Wagon Numbers

作　　者：董咚 王鹏月 Dong Dong;Wang Pengyue(China Energy Railway Equipment Co.Ltd.,Beijing Dongcheng 100120;Beijing jingtianwei Technology Development Co.,Ltd.,Beijing 100085)

机构地区：[1]国能铁路装备有限责任公司,北京东城100120 [2]北京京天威科技发展有限公司,北京100085

出　　处：《现代科学仪器》2022年第1期180-184,共5页Modern Scientific Instruments

摘　　要：目前货车检修正逐步从计划检修向状态检修过渡,就对货车日常管理提出了更高的要求,而对货车车号识别,是实现对货车进行动态跟踪和管理的重要基础。针对货车车号识别,本文采用深度学习技术,利用YOLO网络模型结构和货车车号字符特点,实现对车号位置的准确检测,克服了光照、污迹、缺失等干扰的影响,同时,利用基于卷积神经网络的分类技术,实现对货车车号的识别。At present,the maintenance of railway wagon is gradually transiting from planned maintenance to state maintenance,which puts forward higher requirements for the daily management of railway wagon,and identification of railway wagon numbers is an important basis for dynamic tracking and management of railway wagon.For Railway wagon numbers recognition,deep learning technology is adopted in this paper to realize accurate detection of vehicle number position by utilizing YOLO network model structure and character characteristics of railway wagon number,and overcome the influence of light,stain and missing.At the same time,the classification technology based on convolution neural network is used to realize the recognition of railway wagon numbers.

关 键 词：货车车号识别 深度学习 卷积神经网络 

分 类 号：U29-3[交通运输工程—交通运输规划与管理]