零件图去噪及骨架提取.pdfVIP

零件图去噪及骨架提取 摘 要 近年来由于计算机在我国迅速的普及，各种工作的自动化如雨后春笋般发展 起来。机械工业也不例外，原来手绘图纸和纸张备份的原始方式早已被计算机软 件绘图和备份所代替。这就涉及到了对原有的重要图纸的计算机备份问题，可一 些图纸由于年头过长，扫描仪的质量等问题，常常会出现很多的噪声和线条断裂、 模糊不清等现象。 本论文正是针对这些问题，采用二值化、边缘检测、平滑去噪、数学形态学 处理等多种办法对原始扫描零件图进行处理，再把各种组合处理的结果进行比 较，尽可能找到最有效的去噪和骨架提取的办法。 关键词：二值化、边缘检测、中值滤波、闭运算、细化 I 零件图去噪及骨架提取 Abstract Recent years ，with the rapid development of computer industry in China, various kinds of works with automatization are developing quickly. Mechanical industry is not an exception, and the old way to draw pictures by hands and backup with blueprints were instead by using computer software to draw pictures and backup. So now we should save another copy of the important sources which had been kept in paper by computer. However, because of the long time usage of the blueprint and the problem of scanner, there always appears some problems, such as too much noise, lines rupture, illegibility and so on. The paper focuses on how to deal with these problems by using binarization, edgecontour, noise elimination, msmorphological to disposal the original accessory pictures by scanner. Then compare with the results of different ways, and finally find the most effective way to remove the noise and extract the framework. Keywords ：binarization, edgecontour, median filter, Close algorith, thinning II 零件图去噪及骨架提取 目 录 第一章 前言 1 1.1 零件图的图像处理软件化的背景和意义1 1.1.1 零件图的图像处理软件化的背景 1 1.1.2 零件图的图像处理软件化的意义 2 1.2 零件图的图像处理软件化的方法和流程2 1.2.1 图像处理技术的概述 2 1.2.2 基本的图像处理技术 3 1.3 本论文的主要内容 4 第二章 零件图的二值化6 2.1 预先设定阈值法 7 2.2 动态自适应的阈值选择法 8 2.3 基于直方图的最佳阈值选择法 8 2.4 最大方差