遥感图像黄河冰封期水体分析与应用研究

发布时间：2018-05-17 00:36

本文选题：TM遥感影像 + 最佳波段组合　；参考：《内蒙古农业大学》2014年硕士论文

【摘要】：本文以乌拉特前旗黄河断面为研究区。首先,根据最佳波段组合的选取原则,利用matlab编程语言实现单波段信息提取和相关系数矩阵、信息熵、最佳指数的统计。经反复比对和分析后,得到乌拉特前旗黄河断面畅流期TM542和冰封期TM543的最佳波段组合。经组合后得到的遥感影像能够最大程度的反映地物的信息,使各类地物间的差异明显,为下一步的遥感影像分类奠定基础。其次,在最佳波段组合影像的基础上,开始对遥感图像进行分类操作,经对比分析K-means算法、马氏距离和最大似然法的分类结果,得到分类效果最好的人机交互式最大似然法的分类效果图。接着制作出地物分类的专题图,计算出冰封期各类地物的面积。通过专题图,我们还可以观察各类地物的分布现状,这些都为防凌工作提供了技术支持。最后,进行遥感图像拼接的操作,它可以解决在不同时间不同角度获取的两幅或者多幅图像的匹配处理问题。本文在现有的配准技术中选取了基于特征点匹配的SIFT算法,并引入可以对误匹配点进行优化的RANSAC算法,以便进一步提高匹配的准确度。匹配结束后通过双向线性内插法,对拼接处的影像进行平滑,加强了视觉效果,得到了图像拼接的全景图。
[Abstract]:This paper takes the Yellow River section of the former Banner of Wulat as the research area. Firstly, according to the selection principle of optimal band combination, the single band information extraction, correlation coefficient matrix, information entropy and best index statistics are realized by matlab programming language. After repeated comparison and analysis, the optimal band combination of TM542 and TM543 in the Yellow River section of Qianqi, Wulat, was obtained. The combined remote sensing images can reflect the information of the ground objects to the greatest extent, which makes the difference between different ground objects obvious, and lays a foundation for the classification of remote sensing images in the next step. Secondly, on the basis of the best band combination image, the classification operation of remote sensing image is started, and the classification results of K-means algorithm, Mahalanobis distance and maximum likelihood method are compared and analyzed. The classification effect map of the best human-computer interactive maximum likelihood method is obtained. Then make the thematic map of the classification of features and calculate the area of all kinds of objects in the ice period. Through the thematic map, we can also observe the distribution of various features, which provide technical support for anti-ice work. Finally, the operation of remote sensing image mosaic is carried out, which can solve the problem of matching two or more images obtained at different time and different angles. In this paper, the SIFT algorithm based on feature point matching is selected from the existing registration techniques, and the RANSAC algorithm which can optimize the mismatch points is introduced in order to further improve the accuracy of matching. After matching, the image is smoothed by bidirectional linear interpolation, the visual effect is enhanced, and the panoramic image of image stitching is obtained.
【学位授予单位】：内蒙古农业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TV875;TV882.1

【参考文献】