EPTA算法在字符识别中应用技术.docVIP

学校代码： 10128学 号： 200520203005 学校代码： 10128 学 号： 200520203005 本科毕业设计说明书 本科毕业设计说明书 （ 题 目： 题 目：EPTA算法在字符识别中的应用 学生姓名：高斌 学 院：信息工程学院 系 别：电子信息工程系 专 业：电子信息工程 班 级：电子06-1 指导教师：韩建峰 讲师 二 〇 一 〇 年 六 月 摘 要 字符识别是采用数字图像处理的方法识别字符的一门技术，它在各领域都得到广泛的应用。在字符识别中最重要的预处理过程——细化。在细化算法中，并行细化算法是一种较好的算法。相对于经典的并行细化算法，改进的增强并行细化算法可以更好的解决两像素图像斜线信息丢失、多余枝杈和冗余像素的问题，具有更高的效率和鲁棒性。 本文首先对字符识别系统中的细化及其实现平台MATLAB简单介绍。然后阐述了经典细化算法中的ZS算法和LW算法以及改进后的EPTA算法原理。通过MATLAB实现这几种算法对字符的细化，经过对比分析出EPTA算法相比经典算法没有多余枝杈的情况下很好的解决了两像素斜线丢失问题，对冗余像素的问题也得到了一定的解决。最后分析得出经典算法细化规整清晰字符图像时斜线信息丢失并不严重，EPTA算法在运算速度较经典算稍慢。 关键词：细化；并行细化算法；EPTA；MATLAB Abstract Character recognition is a technology based on digital image processing to identify the character which have been widely used in various fields. In character recognition, the most important pretreatment process is thinning. In thinning algorithms, Parallel thinning algorithm is a more better method for the classical parallel thinning algorithms, the novel Enhanced Parallel Thinning Algorithm can be solves the problems such as information shrink in slanting line, two pixel line preservation and needless trees. Some experiments show that the proposed algorithm is robust and efficient. This dissertation introduce the thinning in character recognition system and the flat roof to achieve it first. Then expound the principle of the ZS algorithm and the LW algorithm in the classical thinning algorithm and the novel EPTA algorithm. After that we achieve the thinning by these algorithms on MATLAB. Analysis the advantage of the EPTA solves the problems such as information shrink in slanting line, two pixel line preservation and needless trees by comparing the classical algorithm. The ultimate analysis obtains the EPTA algorithm solves the neat and clear character not very obvious, the thinning speed of the EPTA algorithm is slowly than the classical algorithm. Keywords: Thinning; Parallel Thinning Algorithm; EPTA; MATLAB 目 录 TOC \o 1-3 \h \z \