融入空间关系的GMM全色高分辨率遥感影像监督分割方法

发布时间：2018-03-09 22:10

本文选题：高分辨率遥感影像　切入点：高斯混合模型　出处：《电子与信息学报》2017年05期 　论文类型：期刊论文

【摘要】：为了解决高分辨率遥感影像中相同地物目标异质性和空间破碎性增大及不同地物目标的相似性增强所带来的分割新问题,该文提出一种融入空间关系的高斯混合模型(GMM)高分辨遥感影像监督分割方法。该方法首先按分割区域进行监督采样,并通过最小二乘法进行直方图拟合,对影像中的每个类别区域建立GMM用来精确表征高分辨遥感影像每个分割区域复杂的地物光谱特征;然后在GMM的概率测度域融入空间关系,使每个像素的区域所属由该像素邻域窗口内所有像素概率测度共同决定,以刻画高分辨率遥感影像中像素间的空间相关性;最后按照最大概率测度原则完成对高分辨率遥感影像的分割。为了验证文中算法的可行性与有效性分别对合成影像及真实高分辨率遥感影像进行分割实验,并和经典的FCM方法及HMRF-FCM方法进行对比,定量与定性的结果证明了文中方法能够提高分割精度。
[Abstract]:In order to solve the new problem of segmentation caused by the increase of heterogeneity and spatial fragmentation of the same ground objects and the enhancement of similarity of different objects in high-resolution remote sensing images, In this paper, a supervised segmentation method of high resolution remote sensing image based on Gao Si mixed model GMMs is proposed. The supervised sampling is carried out according to the segmentation area, and histogram fitting is carried out by least square method. For each region of the image, GMM is established to accurately represent the complex spectral features of each segmented region of the high-resolution remote sensing image, and then the spatial relationship is incorporated into the probability measure domain of the GMM. The region ownership of each pixel is determined by all pixel probability measures in the pixel neighborhood window to describe the spatial correlation among pixels in the high-resolution remote sensing image. Finally, according to the principle of maximum probability measure, the segmentation of high-resolution remote sensing images is completed. In order to verify the feasibility and validity of the proposed algorithm, experiments are carried out on synthetic images and real high-resolution remote sensing images, respectively. Compared with the classical FCM method and HMRF-FCM method, the quantitative and qualitative results show that the proposed method can improve the segmentation accuracy.
【作者单位】：辽宁工程技术大学矿业技术学院;辽宁工程技术大学测绘与地理科学学院;阜新市国土资源局;
【基金】：辽宁省教育厅一般项目(LJYL036,LJYL012) 教育部高等学校博士学科点专项科研基金(20122121110007)~~
【分类号】：TP751

【相似文献】