基于Tsallis熵差的遥感图像边缘检测方法  

Edge detection method based on Tsallis entropy difference of remote sensing image

作　　者：闫海霞[1,2] 赵晓晖[1] 

机构地区：[1]吉林大学通信工程学院,长春130012 [2]吉林大学电子科学与工程学院,长春130012

出　　处：《计算机应用研究》2009年第9期3598-3600,共3页Application Research of Computers

摘　　要：基于Tsallis熵差和百分位形态变换,提出了一种改进的遥感图像边缘检测方法。该方法通过构造基于百分位形态变换的边缘检测算子和选择不同方向结构元素进行变换来增加图像的边缘信息,并且该方法在百分位形态变换的基础上还改进了百分位变换的评价准则。它利用图像Tsallis熵差来选择边缘检测算子的百位变换值,将选择Tsallis熵差最大的百分位变换作为变换结果。实验结果表明,与传统的基于数学形态学的边缘检测方法相比,该方法可以最大程度上抑制噪声,有效地提高图像的边缘检测效果。This paper proposed an algorithm based on Tsallis entropy difference and percentile morphological transform for remote sensing image edge detecting. This method was based on percentile morphological transform in order to enhance the image edge by selecting different direction＇ s structure, furthermore this method changed the evaluation of ＂p＂ value based on percentile morphological transform by the Tsallis entropy difference, and selected the maximum difference of Tsallis entropy. Experiment results show that, compared with the traditional edge detecting method, this method decreases the noise as possible as it can and enhances image edge.

关 键 词：百分位形态学 边缘检测 Tsallis熵差 遥感图像 

分 类 号：TN911[电子电信—通信与信息系统]