基于图像处理的列车车号识别系统  

Train Number Recognition System Based on Image Processing

作　　者：梁鑫 王博 黄卉 苏国梅 李世豪 王仕禛 刘子绰 瞿良玮 李昊 杨然 Liang Xin

机构地区：[1]新疆工程学院控制工程学院,新疆乌鲁木齐830023

出　　处：《工业控制计算机》2024年第6期117-119,共3页Industrial Control Computer

基　　金：新疆维吾尔自治区大学生创新创业训练计划项目(S202210994011)资助。

摘　　要：铁路货车车号在铁路信息化管理中用于快速准确地判断车辆位置和使用情况等重要信息。传统的车号判别方式主要依靠人力手动抄写记录和运用射频技术进行判别,容易出现对车号判别错误的情况,同时也耗费了大量的人力、物力。车号判别出错将直接影响到列车运行调度指挥和货车实时追踪管理,降低铁路列车运转效率。针对以上问题,设计了基于图像处理的车号识别系统,用于采集车号信息数据,构建实验数据集。首先,使用MobileNet压缩训练模型并提取列车车号特征;其次,采用YOLOv2算法在图像中进行目标检测;最后,采用PID算法控制二自由度云台搭载视觉识别系统追踪列车车号所在位置,并将识别到的列车车号上传至PC上位机。对列车车号识别系统的设计,期望为后续升级改造打下基础。Railway freight car number is used in railway information management to quickly and accurately judge the important information such as vehicle location and usage.The traditional identification of vehicle number mainly relies on manual copying and recording and the use of radio frequency technology,which is easy to make mistakes in the identification of vehicle number,and also consumes a lot of manpower and material resources.Aiming at the above problems,a vehicle number recognition system based on image processing is designed to collect vehicle number information data and construct experimental data set.Firstly,MobileNet is used to compress the training model and extract the train number features.Secondly,YOLOv2 algorithm is used to detect objects in images.Finally,PID algorithm is used to control the two-degree-of-freedom head equipped with a visual recognition system to track the location of the train number,and the identified train number is uploaded to the PC.The design of the train number recognition system is expected to lay a foundation for the subsequent upgrading.

关 键 词：图像处理 车号识别 自动追踪 铁路火车车号 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U29-39[自动化与计算机技术—计算机科学与技术]