摘要
针对启发式云变换只能对一维特征进行变换的缺陷,提出一种基于高斯混合模型的改进云变换方法。通过EM算法和高斯分布的拟合误差求解云模型的数字特征,抽取图像底层概念,从而实现图像分割。通过图像分割实验验证了该算法的有效性。
This paper proposes a new cloud transformation method in order to improve the tranditional cloud transformation which cannot deal with multidimensional data. EM algo- rithm and fitting error of Gaussian mixed model are used to extract cloud concepts which are expressed by the digital characteristics. Image segmentation is realized by the improved men- thod. The image segmentation experiments are used to compare the proposed method with traditional methods. The comparison experiments validate the proposed method.
出处
《武汉大学学报(信息科学版)》
EI
CSCD
北大核心
2013年第10期1163-1166,1183,共5页
Geomatics and Information Science of Wuhan University
基金
中国博士后科学基金资助项目(20110491230)
中国地质大学教育部地理信息系统软件及其应用工程研究中心开放课题资助项目(20111105)
中央高校基本科研业务费专项资金资助项目(CUG120833)
关键词
云变换
高斯混合模型
云模型
图像分割
cloud transformation
Gaussian mixed model
cloud model
image segmenattion