基于共生矩阵的遥感图像面向对象分割  

Object-oriented segmentation of remote sensing image based on co-occurrence matrix

作　　者：武维[1] 李玉霞[1] 童玲[1] 顾行发[1,2] 

机构地区：[1]电子科技大学表空间信息技术研究所,四川成都611731 [2]中国科学院遥感应用研究所,北京100101

出　　处：《计算机工程与设计》2011年第2期596-598,共3页Computer Engineering and Design

基　　金：国家自然科学基金项目(60841006);四川省科技支撑计划基金项目(2008SZ0147);四川省杰出青年基金项目(09ZQ026-058)

摘　　要：为了使通过共生矩阵能得到更合理的纹理特征,实验1研究了开窗大小变化对各纹理特征的影响,从而为计算共生矩阵时的开窗大小选择提供依据。进而按照实验1的研究结果,实验2把纹理特征和光谱特征结合起来对高分辨率遥感影像进行了面向对象的分割,并与不考虑纹理信息的分割结果通过优度实验法进行了对比研究。对比结果表明,考虑了纹理信息的面向对象分割能更真实有效地反映图像中地物目标的整体结构,为进一步有效的地物分类提供保障。To obtain more reasoned texture features, the effect of the window size on texture features quality is studied. From the first experiment, a scientific basis for choosing the optimum window size is presented. Through the second experiment, a texture image segmentation algorithm based on object-oriented which combines texture features and spectral characteristics is presented. The advantage of the method based on co-occurrence matrix is evaluated by comparison of the experimental data with that generally used in the past in case the texture features were neglected. The study conclusion shows that the new method more reliable comprehensive in reflecting the ground objects in the image.

关 键 词：共生矩阵 遥感图像 纹理分割 高分辨率 面向对象 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP751.1[自动化与计算机技术—计算机科学与技术]