【作者】 魏君；

【导师】 刘志鹏；

【作者基本信息】 西安理工大学 ， 包装工程， 2003， 硕士

【摘要】 集装箱是一种综合性的大型周转货箱，集装箱运输是集装运输的最主要方式，采用集装箱运输可以获得巨大的经济效益和社会效益。随着国民经济的发展和世界贸易的扩大，我国现有的集装箱码头和货物集散地的硬件设施和管理水平已远远不能适应目前集装箱大型化、专业化的要求。为了加强集装箱的监督和管理，实现集装箱运营管理的自动化并与国际接轨，本论文提出了一种利用计算机视觉系统对集装箱编号进行识别的方法。该方法考虑到集装箱编号的特点，即集装箱字符由于自然或人为因素可能被严重污染，清晰度差等问题，对字符的分割和识别方法进行了深入的研究，主要完成了以下工作： 1．对含有集装箱字符的图像进行预处理，针对集装箱字符特点提出了字符定位方法，从复杂背景中提取出字符并进行分割。 2．依据字符特征，对字符进行了尺寸归一化和笔划粗细归一化，并提取字符特征作为神经网络的输入。 3．设计字母网络和数字网络分别对集装箱编号中的字母和数字进行识别，输出识别结果。 最后，应用所提出的方法，在Win98平台下完整地实现了集装箱编号智能识别系统。实验结果表明，该方法具有较强的抗干扰能力和快速性，识别率可达95％以上，识别时间小于1秒，具有实际应用的前景。更多还原

【Abstract】 Container is a kind of comprehensive and large-scale packing box. Containerized transportation is the main way to transport. It can obtain enormous economic benefit and social benefit to adopt containerized transportation. With the development of national economy and enlargement of world commerce, hardware facility and management level of container quay and distributing center of goods in our country can not meet the maximized and specialized request already. In order to strengthen the supervision and management of the container and realize the automation of management and in line with international standards, In the paper, we present an approach based on computer vision system to recognize the characters on containers. Considering the characteristic of the characters on containers, characters are distorted owing to the influence of natural and artificial factor, we take a profound research on the general problem of character extraction and recognition. Main work in the thesis is as follows:1. Carry on the pretreatment to the picture with container characters and draw the characters and extract from complex scene.2. According to the properties of characters, A method for normalizing theirsize and thickness of strokes is implemented firstly. Then extract featuresfrom characters as the input of neural network. 3. Design letter network and digital network to discern the letters and digitsrespectively and output the result of discerning.The system has been implemented on Win98 by the author independently , the experimental result shows this method has stronger anti-interference ability and the discerning rate can be more than 95 percent and discerning time is less than one second. It has prospect of practical application, ^更多还原