材料图像目标粘连点定位与分割方法研究  

Concave Spots Localization and Region Segmentation in Fibrous Material Image Based on Chain Codes

作　　者：朱煜[1] 江林佳[1] 肖玉玲[2] 

机构地区：[1]华东理工大学信息科学与工程学院,上海200237 [2]河南工业大学化学工业职业学院机械电子系,河南郑州450042

出　　处：《南京理工大学学报》2008年第1期110-113,共4页Journal of Nanjing University of Science and Technology

基　　金：国家自然科学基金(50403005);华东理工大学青年骨干教师资助项目(0156101)

摘　　要：该文所处理的纤维材料图像由扫描电子显微镜获取。图像具有目标孔洞粘连明显、重叠、彩色信息较弱的特点。通过对图像中亮度V(Value)分量进行操作,利用方向链码可以表示曲线率的特点,计算出链码差,从而提取出目标粘连处的特征凹点,并通过凹点之间的相关性,采用局部曲率最大的原则对提取的凹点进行筛选,在此基础上实现了对凹点的定位和较为成功的目标区域分割,解决了后续参数提取工作的关键问题。The fibrous material images are captured from a scanning electronic microscopy （SEM）. The objects within the image overlap, join together and possess weak chromatic information. The operation is implemented for V component of the image. A concave spots localization method is proposed. The concave pixels are achieved by the information of curvature and obvious direction change, which is calculated by chain code difference of three pixels. An optimized algorithm based on local maximum curvature is used to extract concave spots and to match them. A geometrical region segmentation method is proposed and the objects parameter extraction is discussed. Experimental results show that the algorithm proposed performs well.

关 键 词：材料图像 链码 凹点定位 区域分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]