基于区域相似性的高分辨率遥感影像分割  

High resolution remote sensing image segmentation based on region similarity

作　　者：赵泉华[1] 谷玲霄 李玉[1] Zhao Quanhua;Gu Lingxiao;Li Yu(Institute for Remote Sensing Science and Application, School of Mapping and Geographical Science, Liaoning Technical University, Fuxin 123000, China)

机构地区：[1]辽宁工程技术大学测绘与地理科学学院遥感科学与应用研究所

出　　处：《仪器仪表学报》2018年第2期257-264,共8页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(41301479);辽宁省自然科学基金(2015020090)项目资助

摘　　要：作为蕴含于高分辨率遥感影像中的重要信息,地物目标的光谱特征和纹理结构对其精准分割至关重要。为此,结合高分辨率遥感影像的光谱和纹理信息,提出基于区域相似性的高分辨率遥感影像分割算法。首先,将图像域划分为一系列同质子区域;然后,在区域基础上,结合定义的纹理相似性和光谱相似性,以获取区域相似性;再利用基于区域相似性的分形网络演化算法（FNEA）实现高分辨率遥感影像分割。利用所提出算法,分别对合成纹理图像和高分辨率遥感影像进行分割实验,实验结果表明所提出算法具有一定的可行性及有效性。As the important information contained in high resolution remote sensing image,the spectral characteristic and texture structure of physical object are very critical to its accurate segmentation. Therefore,combining the spectral and texture information of high resolution remote sensing image,a segmentation algorithm of high resolution remote sensing image based on region similarity is proposed.First of all,over segmentation is used to divide the image into a series of homogeneous sub-regions. Then,the region similarity is obtained on the regional basis combining texture similarity and spectral similarity. After that,the Fractal Net Evolution Approach（ FNEA） algorithm based on region similarity is adopted to accomplish high resolution remote sensing image segmentation. The proposed algorithm was used to conduct experiment on synthetic texture image and high resolution remote sensing images,the experiment results show the feasibility and effectiveness of the proposed algorithm.

关 键 词：高分辨率遥感影像分割 过分割 区域相似性 分形网络演化算法 

分 类 号：TH761[机械工程—仪器科学与技术]