融合色调统计特征的自适应区域增长算法  

Adaptive region growing algorithm combined hue statistical characters

作　　者：柳佳雯[1] 梁光明[2] 刘任任[1] 谢雅婷[1] 丁建文 

机构地区：[1]湘潭大学信息工程学院,湖南湘潭411105 [2]国防科技大学电子科学与工程学院,长沙410073 [3]长沙爱威科技,长沙410013

出　　处：《计算机应用研究》2012年第8期3176-3178,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60673193)

摘　　要：针对区域增长算法种子点选取及生长条件对分割效果影响大,但难以有效、准确确定的缺点,提出了一种融合色调、色调频度及其曲率的自适应区域增长算法(HARG)。通过曲率求取分割对象的色调频度曲线峰值,以峰值为中心进行加窗累加,当和值满足门限时,峰值作为区域增长的种子点,门限作为区域增长的生长条件。将算法应用于显微镜细胞图像的单个细胞内部分割,实验结果表明,HARG算法自适应确定了种子点及生长条件,能够实现多区域分割,分割精度得到改善。This paper proposed a adaptive region growing algorithm （HARG） combined hue, hue frequency and its discrete curvature, to direct at the drawback of conventional region growing algorithm characterized by the seed selected and growing condition confirmed great effected and ineffective confirmed. It calculated seed using the sum of hue histogram added window fulfilled with threshold, and region growing by the threshold. Used the algorithm for single cell of microscope cell image, and the experiment results show that the HARG algorithm enhances segmentation precision and achieves segmentation of multiple regions by ensuring adaptively the seed and growing condition.

关 键 词：彩色图像分割 加窗色调直方图 离散曲率 区域增长算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]