融入空间关系的二型模糊模型高分辨率遥感影像分割

发布时间：2018-06-25 09:10

本文选题：区间二型模糊模型 + 不确定轨迹　；参考：《遥感学报》2016年01期

【摘要】：为解决高分辨率遥感影像分割中,由光谱测度的空间复杂性、相同类型地物目标异质性增大带来的类属不确定性以及分割决策不确定性等引起的分割精度下降问题,提出一种融入空间关系的区间二型模糊模型高分辨率遥感影像监督分割方法。(1)建立高斯函数模型作为一型模糊模型,用来刻画像素类属的不确定性;(2)模糊化一型模糊模型中的均值或标准差,建立区间二型模糊模型,以强化类属的不确定表达和增加分割决策信息;(3)综合一型模糊模型及区间二型模糊模型的上、下隶属函数建模模糊决策模型;(4)融入邻域像素关系,使用待分像素及其邻域像素在模糊决策模型中的隶属度共同决定像素的类属。采用本文算法分别对真实高分辨遥感影像及合成影像进行分割,并对测试结果进行定性和定量分析。结果表明,本文算法可以得到更高的分割精度。
[Abstract]:In order to solve the problem of reducing the segmentation accuracy caused by the spatial complexity of spectral measure, the uncertainty of classification and the uncertainty of segmentation decision caused by the increasing heterogeneity of objects of the same type of ground objects, the segmentation accuracy of high resolution remote sensing image segmentation is reduced. An interval type 2 fuzzy model with spatial relationship is proposed. (1) the Gao Si function model is established as a fuzzy model. It is used to describe the uncertainty of pixel genus. (2) the mean value or standard deviation of fuzzy model is fuzzied, and the interval type 2 fuzzy model is established. In order to enhance the uncertain expression of the genus and increase the segmentation decision information, (3) integrating the upper and lower membership function of the fuzzy model of type one and the fuzzy model of interval type two, the fuzzy decision model is modeled; (4) the fuzzy decision model is integrated into the neighborhood pixel relationship. The membership degree of pixels to be divided and their adjacent pixels in the fuzzy decision model is used to determine the class of pixels. In this paper, the real high resolution remote sensing image and synthetic image are segmented, and the test results are analyzed qualitatively and quantitatively. The results show that the proposed algorithm can achieve higher 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;
【基金】：教育部高等学校博士学科点专项科研基金资助(编号:20122121110007) 国家自然科学基金面上项目(编号:41271435)~~
【分类号】：P237

【相似文献】