【作者】 赵有婷； 余志； 李熙莹； 张格格； 丁卉；

【Author】 Zhao Youting;Yu Zhi;Li Xiying;Zhang Gege;Ding Hui;Research Center of Intelligent Transportation System,Sun Yat-sen University//Guangdong Provincial Key Laboratory of Intelligent Transportation System;

【机构】 中山大学智能交通研究中心//广东省智能交通系统重点实验室；

【摘要】 车牌识别系统的评测和图像库的搭建是评价和推动车牌识别技术的关键因素,具有重要的意义。本文在现有车牌识别技术的基础上制定了车牌识别系统评测协议,包括评测指标和评测方案设计;并且介绍了SYSU-ITS系统功能评测车牌图像库,包含6588张各种类型和各省、自治区、直辖市车牌图像;最后在评测协议和图像库的基础上对五个国内常见车牌识别商业系统做了功能评测,评测结果表明目前国内车牌识别系统的瓶颈在于部分车牌种类的定位和复杂汉字的识别。通过研究表明本文制定的车牌识别系统评测协议以及建立的系统功能评测图像库可以有效地评价系统的功能识别差距,并指导国内车牌识别领域未来的研究方向。更多还原

【Abstract】 Two of the most significant factors in support of the development of license plate recognition are a testing procedure to evaluate systems and a standard database of license plate images.In this paper,on the basis of the existing license plate recognition technologies,a standard test protocol are established including evaluation metrics and test design.The SYSU-ITS license plate images database for system function evaluation is established.To date,6588 images are included in the database,which consists of type variation and provincial abbreviation variation images.Finally the experimental results from the 5 license plate recognition systems test are reported.Evaluation results show that the bottleneck of the license plate recognition technology is the location of the some types of license plate and the identification of complex Chinese characters.The research shows that the evaluation methodology and the SYSU-ITS database can effectively evaluate the systems function and identify future directions of research.更多还原