【作者】 黄丽华；

【导师】 王家平；

【作者基本信息】 浙江大学 ， 机械制造及其自动化， 2004， 硕士

【摘要】 自20世纪60年代末第一台图像分析仪诞生以来，人们渐渐地走出传统的辛苦而又低效的人工金相分析，开始利用计算机来进行计算机辅助定量金相分析，因其与传统的人工金相分析相比具有处理速度快、数据精确可靠、实验结果的重现性好等一系列的优点，已经逐渐地取代传统的金相分析方法。 不过，纵观这三、四十年计算机辅助定量金相分析的发展，国内大多是停留在黑白金相图像的处理上，即利用灰度来进行图像的处理和分析。随着对金相定量要求和精度的提高，这种基于灰度的黑白图像分析在许多定量金相的分析任务中已经显出有些无力。随着彩色图像处理技术的快速发展和计算机储存与处理能力的飞速提高，目前许多国家研究人员已经着手研究利用彩色金相图像来进行分析，因为其衬度鲜明、可分辨能力强等特点。 本课题是南京东图数码科技公司(即原江南光学仪器厂的技术开发部门)的一个委托研究项目，在该公司提供的实验设备的基础上，建立一套彩色金相计算机辅助分析系统(CMAS-Color Metallography Image Compute-aided Analysis System)。该系统集XJL—02立式显微镜、CCD摄像头、视频图像采集卡、计算机显示系统于一体，实现了从彩色金相图像记录、采集、预处理到特征参数测量、结果分析和打印等一系列功能。 该系统针对彩色金相显微图像的特点，重点建立一系列适合金相应用的彩色图像处理方法作为CMAS系统的重要组成部分。这些方法包括彩色增强(色调／饱和度调整、亮度／对比度调整、多通道颜色调整)，颜色变换(彩色图像灰度化、真彩色图像256色化)，矢量滤波，和作重点研究的彩色图像分割(聚类法、阈值法、区域生长法、边缘提取)。更多还原

【Abstract】 Compared with traditional quantitative analysis of metallography (QAM), computer-aided QAM has more advantages such as faster speed, high efficiency, more credible, reproducibility easily. With the development of computer technology, it is inevitably that the computer-aided QAM will take the place of the traditional QAM.But, in recently two or three decades in China, the research on computer-aided QAM stays on the grey; this is insufficiency in some quantitative metallography assignment because of the improvement of the QAM requirement and precision. With the rapidly development of color image processing and the computer storage, in many academe color metallography is being used in QAM to get more information because of its intensive contrast and strong discrimination.We developed the Color Metallography Image Compute-aided Analysis System (CMAS). The System is composed of an XJL-02 standing microscopy, a CCD camera system, a video image sampling card, a computer system and the software what we developed specially for this system. Furthermore, it provides many powerful functions including image recording, image sampling, image processing, parameter measurement, result analysis and so on.According to the characters of color metallography image, we implemented various kinds of color image processing approaches, including color enhancement (Hue/Saturation adjustment, Brightness/Contrast adjustment, multi-channel color adjustment), color transform (color to gray, true color to 256 colors), vector filtering, and most of important, color image segmentation (clustering, threshold, region growing, edge detection). All of these are the important parts of CMAS.更多还原