【作者】 张海军；

【导师】 王春光；

【作者基本信息】 内蒙古农业大学 ， 农业机械化工程， 2008， 硕士

【摘要】 本课题选取由石墨和不同基体组织组成的灰铸铁作为研究材料,根据定量检测标准选取放大倍数为100×、200×和500×进行金相图像的采集。利用数字图像处理技术对采集到的金相图像进行处理,实现灰铸铁金相分析。并得到以下成果和结论:1.针对灰铸铁金相图像出现的亮度不均匀现象,利用空域阴影校正法对金相图像进行亮度不均匀校正;利用小波变换与均值滤波相结合的算法对灰铸铁金相图像进行去噪处理,并取得了较好的结果。2.利用常用的分割算法:阈值化法、FCM聚类法、遗传算法和小波变换对灰铸铁金相图像进行分割,分析了分割效果和算法的性能。提出了一种适合于灰铸铁金相图像的分割算法:在小波域进行的基于二维最大类间方差的遗传算法,提高了分割效率,并得到了较好的分割效果。3.利用面积法实现了石墨数量、铁素体数量以及珠光体数量的计算;利用像素间距离实现了石墨长度的测量;利用分形数学的思想实现了灰铸铁石墨形状的分类;利用两点间距离公式实现了珠光体片间距离的测量。4.基于Matlab7.0,开发了一种灰铸铁金相分析系统,完成特征参数的测定。更多还原

【Abstract】 Grey cast iron was selected as the researched materials in this paper. According to the standard of quantitatively detecting, experiment selected enlarge multiple is 100×, 200×and 500×to capture metallographic image. This paper provided the study on microstructure image processing and analyzing methods based on digital image processing. The main achievements and conclusions are as follows:1. Aiming at the non-uniform phenomenon of microstructure image brightness, one method which based on the theory of shadow remedy was used to remedy it; the algorithms which combined wavelet transform and mean filtering was used to reduce the noise of image.2. Some segmentation methods, such as threshold method, FCM method, Genetic algorithm and wavelet transform, were used to segment the grey cast iron microstructure image, and this paper analyzed the segmentation results and performance of algorithm. In this foundation, an approach for grey cast iron image segmentation, called Genetic algorithm based on 2-D maximum between-cluster variance in wavelet domain, was presented. Segmentation result shows that the algorithm achieves good performance.3. Graphite quantity, ferrite quantity, pearlite quantity were calculated by area, graphite length was calculated by the distance of two pixel points, graphite form of grey cast iron was recognized by fractal theory, the span between pearlite flakes was calculated by the distance of two pixel points in this paper.4. A analyzing system of grey cast iron metallography was developed based on Matlab.更多还原