基于“催化剂”的图像分割算法研究  

Image segmentation based on "catalyst" segmentation method

作　　者：康庄[1] 邓晓刚[2] 吕庆文[2] 

机构地区：[1]中山大学第三附属医院,广州市510630 [2]南方医科大学,广州市510515

出　　处：《医疗卫生装备》2007年第11期3-5,共3页Chinese Medical Equipment Journal

基　　金：国家"973"重点基础研究发展规划项目基金(2003CB716101);广东省自然科学基金(06024419)资助

摘　　要：目的:针对基于二叉树分割算法的分割停止判断的局限性,提出基于"催化剂"的图像分割方法。方法:在分割图像中引入试探图像作为"催化剂,"并以"催化剂"在分割结果中是否显现为依据来判断二叉树分割的候选结点是否应该停止分割。结果:"催化剂"分割算法可有效地对待分割图像进行正确的类数判断。结论:初步实验表明,"催化剂"分割方法简洁、有效。Objective To propose a novel method for image segmentation hased on ＂catalyst＂ segmentation method. Methods Feeler image was imported into segmentation algorithm as the ＂catalyst＂, and whether the candidate nodes should stop segmentation or not was decided according to whether the catalyst appeared in the segmentation result. Results Image segmentation based on ＂catalyst＂ segmentation method could judge the right class number for segmented images. Conclusions Primary experiments shows that ＂catalyst＂ segmentation method is simple and useful.

关 键 词：图像分割 催化剂 二叉树 马尔可夫随机场 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]