多分辨模型下的无监督统计纹理分割算法  

Unsupervised statistical texture segmentation based on multi-resolution model

作　　者：周晖[1] 王润生[1] 

机构地区：[1]国防科学技术大学ATR重点实验室,湖南长沙410073

出　　处：《计算机应用》2006年第1期129-131,共3页journal of Computer Applications

摘　　要：针对基于统计的纹理分割算法存在的不足,提出了一种新的多分辨模型下的无监督统计纹理分割算法。该算法分层次对纹理图像进行动态分割,有效地利用了不同分辨率上最能表述某一纹理特性的统计特征,然后结合纹理的结构信息对边缘区域进行边界提取,进而得到较准确的纹理分割图。实验证明了这种区域与边界相结合的算法对纹理图像的分割具有高效、准确、通用性强的特点。In order to overcome the weakness of statistical texture segmentation method, a new unsupervised texture segment algorithm, based on multi-resolution statistical model, was present. It first found the best feature that explained the texture clearly in different resolution and then the dynamic cluster on different level. At last, by combining the structure information of texture edge, the edge of different patterns was extracted to get a relatively accurate result of texture segmentation. The results show that this algorithm based on region and edge is an efficient, accurate and robust method of texture segmentation.

关 键 词：纹理分割 多分辨率 动态聚类 边缘提取 无监督 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]