基于云模糊理论的图像纹理分割  

Image Texture Segmentation Based on Cloud Fuzzy Theory

作　　者：王佐成[1] 李永树[2] 薛丽霞[1] 汪林林[1] 

机构地区：[1]重庆邮电大学软件学院,重庆400065 [2]西南交通大学土木工程学院,四川成都610031

出　　处：《西南交通大学学报》2007年第5期548-552,共5页Journal of Southwest Jiaotong University

基　　金：重庆市教育委员会科学技术研究项目资助(KJ060511)

摘　　要：为了处理图像纹理的模糊性和随机性,基于云模糊理论提出了纹理特征矢量云模型,并成功地应用于纹理图像分割.该方法在对纹理统计描述符模糊化处理后,逆向生成纹理特征矢量云.矢量云模型的数字特征能够很好地表达纹理的模糊性和随机性,据此通过云距离计算及纹理特征矢量云生长,完成对图像纹理的分割.实验结果表明,该方法较经典的ISODATA算法和K-means簇算法的分割精度高,并且迭代收敛速度快.To deal with the fuzziness and randomness of image textures, a vector cloud model for texture features was proposed on the basis of the cloud fuzzy theory and was successfully used to the segmentation of textural images. With this method, texture feature vector cloud is generated backward after fuzzily processing the statistical descriptor of textures. The fuzziness and randomness of textures can be represented perfectly by the digital character/stics of the vector cloud model, and as a result, image segmentation can be accomplished by Cloud distance calculating and texture feature vector cloud growing. The research result shows that the segmentation accuracy is higher and the convergence speed is quicker algorithm using the proposed model than using the iterative self-organizing data analysis techniques （SODATA） and the K-means cluster algorithm

关 键 词：纹理分割 云理论 纹理特征矢量云 纹理统计描述符 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]