摘要
针对基于粗糙熵的图像分割算法不能满足复杂图像的多类目标提取的需要,本文先利用K-均值聚类算法对图像进行区域分割,再利用基于粗糙熵的方法对分割结果进行目标提取,从而达到多阈值分割的目的。通过对遥感图像进行分割处理,证明了改进后算法的有效性。
An image segmentation algorithm is based on ＂rough entropy＂ and its computational complexity is linear. But as a method of single threshold segmentation, it can not satisfy the need for multiclass segmentation of complex image. Therefore, K-means clustering algorithm is used to segment the image based on region firstly, And then the algorithm based on ＂rough entropy＂ is used to extract objects from the segmentation results, so as to achieve multi-threshold segmentation. The experimental results of remote sensing image segmentation demonstrate that the improved approach is valid.
出处
《华东理工大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2007年第2期255-258,共4页
Journal of East China University of Science and Technology
基金
国家自然科学基金项目(60273043)
安徽省高校拔尖人才基金项目(05025102)
安徽省自然科学基金项目(050420204)
关键词
粗糙熵
K-均值聚类
图像分割
rough entropy
K-means clustering
image segmentation
