【作者】 卢盛辉；

【导师】 唐慧佳；

【作者基本信息】 西南交通大学 ， 计算机应用技术， 2003， 硕士

【摘要】 集装箱自动识别系统在海关物流监控系统、港务局集装箱管理系统、场站集装箱管理系统、加工区监管系统、运输行业集装箱管理系统等领域有着广泛的应用。基于光学字符识别技术的集装箱自动识别系统主要是通过图像处理技术，自动提取集装箱箱号并识别，由于其具有不需要在集装箱上增加任何其他设备以及可通过人机界面，加入人工干预提高系统性能的优点，发展较快，日益受到人们的重视。 本文研究的是基于光学字符识别技术的集装箱自动识别系统，简要介绍了系统的工作原理，分析了集装箱原始图像的特点，详细阐述了系统处理流程：图像预处理、箱号字符区域定位、箱号字符区域二值化及字符分割、识别预处理、特征提取和字符识别。本文运用形态学运算消除背景、进行灰度变换增强对比度和进行平滑过滤噪声来完成图像的预处理。对预处理后的图像进行边缘提取，先消除非字符边缘的长线条，再用形态学运算消除干扰边缘和噪声。针对字符排列相对集中的特点，详细介绍了如何利用边缘结合形态学定位箱号字符的方法来实现箱号字符的定位。选用合适的二值化方法完成字符区域的灰度图像到二值图像的转换。利用基于求连通区域与投影相结合的字符切分技术完成字符的分割。讨论了字符识别技术，实现了利用图像矩特征进行字符识别以及利用数字的结构特征识别数字。 实验表明，文中的箱号定位方法是有效的，字符识别方法亦具有一定的参考价值。更多还原

【Abstract】 Container Automatic Equipment Identification (AEI) system can be widely applied to material circulation monitoring system in customs, to container management system in harbor bureau, container depot, and transportation industry, and to monitoring and management system in machining center, etc. Container AEI system based on Optical Character Recognition (OCR) technology makes use of image processing technology to extract and recognize characters of container codes automatically. On account of the advantages that such system doesn’t need to attach any other equipment on container when recognizing, and its performance can be improved by means of manual operation through the user interface of computer terminal, it is developed rapidly and attached great importance by people increasingly.In this thesis, the container AEI system based on OCR technology is investigated. First, the working principle of the system is introduced and the characteristics of container’s original image is analyzed. Then the process procedures are discussed in detail, which include image preproccss, localization and threshold of container codes region, characters segmentation, recognition preproccss, feature extraction and recognition. In image preprocessing, image quality is improved by using mathematical morphology to remove background of the image, making gray-scale transformation to enhance contrast and smoothing processing to filter noise. After image preprocessed, its edge is detected, and some long lines in edge image are filtered, and then mathematical morphology is employed to remove edge noise. Considering the characteristic of characters arrayed concentrated, the method that how to make use of character edge and mathematical morphology to locate container codes is described. The gray image in characters area is converted to binary one by suitable threshold. The segmentation technology that combines the technologies of getting connected region and projecting can be used to segment characters. Finally, characters are recognized by utilizing moments to recognize alphabetic character and structure information of numeric character to recognize number.The experiments’ results show that the method of locating container codes is effective and the method of character recognition has some reference value.更多还原