数字图像信息读取与识别模型建模
数字图像信息读取与识别模型建模(任务书,开题报告,外文翻译,论文13000字)
摘要
数字图像的处理和识别，在人们的日常生活中的作用越来越大。本次论文简介了数字图像处理技术、国内外的发展现状、数字图像中的主要技术。着重描述了图像处理技术中的模式识别技术，图像识别包括光学字符识别和生物特征识别。本文用到的是OCR技术，简单介绍OCR技术，手写数字识别属于OCR技术。介绍手写数字识别的目的与意义。介绍了对手写数字识别系统的设计的基本流程结构。研究了OCR现在的发展状况，了解了几款OCR识别软件，如Readiris Corporate等，验证了现如今OCR识别的准确度和速度，实验了OCR技术的处理过程。阐述了识别过程预处理提取感兴趣区域（ROI），特征提取中获取数字各区域的像素值，分类识别中采用了模版匹配法，使用的主要思想是KNN算法，计算目标与模版的距离进行分类，并计算出识别准确率。分类识别中的神经网络和支持向量机都很使用，识别度高，前者支持大规模，后者在两类中发挥很好。介绍了编程时使用的主要工具Opencv。对该工具有简单的了解，了解Opencv中图片的数据结构，编程中常用的函数，详细地阐述了设计的过程，软件的使用配置和程序的编写，通过VS与Opencv的结合使用，调用Opencv中的函数，编写了一个手写数字识别系统。本系统实现的主要功能就是通过手写板写出一个数字后，系统通过已有样本的分类识别出来。实验中，手写了0到9数字，然后通过识别判断是否属实。实验结果中，一共有两个数字发生错误。分析结果得出，本次试验中训练测试样本数量较少，导致识别的结果不够准确。后续改进应该是增加样本数目或者联机识别。
[资料来源：http://doc163.com]
关键字：模式识别；Opencv；手写数字
Abstract
Digital image processing and recognition, in people's daily life, the role of more and more. This paper introduces the digital image processing technology, the development status of the domestic and foreign, the main technology of digital image. The pattern recognition technology in image processing technology is emphatically described, including optical character recognition and biometric identification. This paper uses the OCR technology, a brief introduction of OCR technology, handwritten numeral recognition is OCR technology. The purpose and significance of handwritten numeral recognition is introduced. This paper introduces the basic flow structure of the design of the handwritten numeral recognition system. Research on the development of OCR now, understand several OCR recognition software, such as Corporate Readiris, verify the accuracy and speed of the current OCR recognition, the experimental OCR technology processing. It expounds the recognition preprocessing to extract the region of interest (ROI), the feature extraction of obtaining digital pixel values, classification and recognition of the template matching method, and use the main idea is the KNN algorithm, calculating the distance of the target and template are classified, and the recognition accuracy rate is calculated. Classification recognition in the neural network and support vector machines are used, the recognition is high, the former support a large scale, the latter in the two class to play a very good. This paper introduces the main tools used in programming Opencv. Of the tool is simple to understand, understand the opencv image data structure, programming of commonly used functions, detailed describes the process of design, software configuration and program written, through the use of a combination of VS and opencv, call opencv function, the preparation of the a handwritten numeral recognition system. The main function of this system is to write a number through the tablet, the system through the classification of the existing samples identified. In the experiment, handwritten 0 to 9 numbers, and then through the recognition to determine whether it is true. The results of the experiment, a total of two digital error occurred. The analysis results show that the number of training samples is small in this experiment, and the result is not accurate enough. Follow up improvement should be to increase the number of samples or on line identification.
[版权所有：http://DOC163.com]
Keywords：pattern recognition;Opencv; handwritten
[资料来源：https://www.doc163.com]
目录
摘要 I
Abstract II
第一章绪论 1
1.1 研究背景及意义 1
1.2 国内外研究现状 1
1.3 研究目的 2
1.4 目标识别系统的设计 3
1.5 本文主要工作和文章结构 3
1.6 本章小结 4
第二章数字识别系统的理论研究 5
2.1 获取数字图像 5
2.2 图像预处理 5
2.3 特征提取 5
2.3.1 颜色特征 6
2.3.2 纹理特征 6
2.3.3 形状特征 7
2.3.4 空间关系特征 7
2.4 模式识别 7
2.4.1 模版匹配法 7
2.4.2 神经网络法 8
2.4.3 支持向量机分类法 8
2.4.4 最邻近分类法 8
2.5 本章小结 9
第三章手写数字识别系统的设计 11
3.1 Opencv简介 11
3.2 Opencv的配置 12
3.3 Opencv中常用的函数及功能 15
3.4 手写数字识别系统的设计 15
3.4.1 图像获取 16
3.4.1 预处理 16
3.4.2 特征提取 18
3.4.3 分类识别 18
3.5 实验结果分析 19
3.5本章小结 24
第四章工作总结以及对未来的展望 25
参考文献 26 [来源：http://Doc163.com]
致谢 27