基于MSR-cut的高空间分辨率遥感影像边缘检测分割  

Edge Detection Segmentation Method for High Spatial Resolution Remote Sensing Image Based on MSR-cut

作　　者：杨泽楠 牛海鹏[1,2] 黄亮 王枭轩[1] 刘轲 YANG Zenan;NIU Haipeng;HUANG Liang;WANG Xiaoxuan;LIU Ke(School of Surveying and Land Information Engineering,Henan Polytechnic University,Jiaozuo 454005,China;Research Centre of Arable Land Protection and Urban-rural High-quality Development of Yellow River Basin,Jiaozuo 454005,China;Faculty of Land Resources Engineering,Kunming University of Science and Technology,Kunming 650093,China;Surveying and Mapping Geo-Informatics Technology Research Center on Plateau Mountains of Yunnan Higher Education,Kunming 650093,China;Institute of Remote Sensing Application,Sichuan Academy of Agricultural Sciences,Chengdu 610066,China)

机构地区：[1]河南理工大学测绘与国土信息工程学院,焦作454005 [2]黄河流域耕地保护与城乡高质量发展研究中心,焦作454005 [3]昆明理工大学国土资源工程学院,昆明650093 [4]云南省高校高原山区空间信息测绘技术应用工程研究中心,昆明650093 [5]四川省农业科学院遥感应用研究所,成都610066

出　　处：《农业机械学报》2021年第8期154-162,共9页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金项目(41371524);四川省应用基础研究(面上)项目(2017JY0284);四川省省院省校合作(重点)项目(2018JZ0054)。

摘　　要：针对R-cut(Ratio cut)边缘检测分割模型对高分辨率遥感影像分割时存在过分割和模糊边缘敏感性问题,提出了一种多尺度R-cut(Multi-scale ratio cut,MSR-cut)的遥感影像边缘检测分割方法。首先,采用形态重建的分水岭分割算法对影像过分割,形成多个超像素区域;然后计算并提取影像各个区域的纹理特征信息熵值、光谱特征与邻域均值差分归一化值,分别进行同质性和异质性的有效衡量;并构建评价函数获取最优分割尺度,对这些超像素区域进行初步合并,得到影像的粗分割结果;最后结合各地物的边界权重信息,从全局角度用R-cut的方法对粗分割结果进一步合并,完成对影像的精细分割,生成最终的分割结果。实验选取5个不同场景的高分辨率遥感影像,采用定性和定量两种方法对比分析本文方法与传统R-cut边缘检测分割、Spectral-Rcut边缘检测分割和Textured-Rcut边缘检测分割方法。实验结果表明,MSR-cut边缘检测分割方法能够有效提高分割精度,增强噪声鲁棒性,可取得较好的分割视觉效果。Aiming at the over-segmentation and fuzzy edge sensitivity problems of the R-cut(Ratio cut)edge detection segmentation model for high-resolution remote sensing image segmentation,a multi-scale R-cut(Multi-scale ratio cut,MSR-cut)method was proposed for remote sensing image edge detection and segmentation.Firstly,the watershed segmentation algorithm of morphological reconstruction was used to over-segment the image to form multiple super-pixel regions;then the texture feature information entropy value,spectral feature and neighborhood mean difference normalized value of each region of the image were calculated and extracted,and the same was performed respectively.Effective measurement of qualitative and heterogeneity was made;and an evaluation function was constructed to obtain the optimal segmentation scale,and initially merged these super-pixel regions to obtain the rough segmentation result of the image;finally,combined the boundary weight information of various objects,and R was used from a global perspective.The R-cut method further merged the coarse segmentation results,the fine segmentation of the image was completed,and the final segmentation result was generated.The experiment selected high-resolution remote sensing images of different scenes,and the method was compared and analyzed with traditional R-cut edge detection segmentation,Spectral-Rcut edge detection segmentation and Textured-Rcut edge detection segmentation methods by using qualitative and quantitative methods.Experimental results showed that the MSR-cut edge detection segmentation method can effectively improve segmentation accuracy,enhance noise robustness,and achieve better segmentation visual effects.

关 键 词：高空间分辨率 遥感影像 边缘检测分割 多尺度R-cut 

分 类 号：P237[天文地球—摄影测量与遥感]