基于遥感影像滑坡边界自动提取方法的研究

发布时间：2018-03-30 21:49

本文选题：滑坡　切入点：区域生长　出处：《西南交通大学》2013年硕士论文

【摘要】：滑坡是一种很严重的地质灾害现象,而地震是滑坡产生的直接诱因之一。滑坡的产生通常会引起人员伤亡和经济损失,较大的地震引起的滑坡常常会使得房屋掩埋,道路损坏以及堰塞湖的形成,特别是道路的损坏直接会使得救援通道的阻塞,影响救援速度。快速提取滑坡边界,估算滑坡致灾面积,推测出滑坡的规模,对抗震救灾的快速应急响应提供了有效的数据支持。从遥感影像中提取滑坡特征目前已经有很多研究,但是仍然存在很多问题,这是由于滑坡影像特征的复杂性和多样性。本文围绕着滑坡边界的快速提取开展研究,通过对近几年来滑坡信息提取方法的分析和研究,探讨了滑坡边界提取的基本思想,对现有的图像分割方法加以改进。针对不同的影像特征,本文采用了三种方法对滑坡边界进行提取。首先对地震产生的新滑坡影像特征进行探讨,针对滑坡影像的特定性质采用了基于区域生长、基于边缘区域生长和基于K-均值聚类法这三种方法对滑坡进行边缘提取。基于区域生长的方法中,根据滑坡影像的色彩特征,引入颜色相似性度量作为生长的准则,将颜色相近的滑坡区域合并,再根据边缘提取方法提取滑坡边缘,对边缘进行平滑、细化获得边界,最后对闭合的滑坡边界进行填充估算出滑坡致灾面积。基于边缘区域生长的方法是先对全局图像进行边缘提取,通过区域生长来提取滑坡区域,最后得到滑坡的边界和致灾面积。基于K-均值聚类法是利用K-均值聚类法对图像进行分割,对分割出来的图像进行边缘提取,通过区域生长方法获得滑坡边缘,最后量算滑坡致灾面积。本文针对提出的算法编写了相应的程序,包括图像平滑、图像分割、边缘处理、图像填充等功能,较好的实现了滑坡边界提取和滑坡致灾面积估算功能。
[Abstract]:Landslide is a very serious geological disaster phenomenon, and earthquake is one of the direct inducements of landslide. Landslide usually causes casualties and economic losses, and landslides caused by larger earthquakes often make houses buried. The road damage and the formation of the barrier lake, especially the damage of the road, will directly block the rescue passage and affect the rescue speed. Quickly extract the landslide boundary, estimate the landslide disaster area, and speculate the scale of the landslide. It provides effective data support for rapid emergency response to earthquake relief. There have been many researches on landslide feature extraction from remote sensing images, but there are still many problems, which is due to the complexity and diversity of landslide image features. Based on the analysis and study of landslide information extraction methods in recent years, the basic idea of landslide boundary extraction is discussed, and the existing image segmentation methods are improved. In this paper, three methods are used to extract the landslide boundary. Firstly, the characteristics of the new landslide image produced by earthquake are discussed, and the region based growth is adopted in view of the specific properties of the landslide image. The edge of landslide is extracted based on the three methods of edge region growth and K-means clustering. In the method of region growth, according to the color characteristics of landslide image, the color similarity measure is introduced as the growth criterion. Combining the similar color landslide area, then extracting the landslide edge according to the edge extraction method, smoothing the edge, thinning the obtained boundary, Finally, the landslide disaster area is estimated by filling the closed landslide boundary. The method based on the growth of the edge region is to extract the edge of the global image first, and then extract the landslide area by the region growth. Finally, the boundary and disaster area of landslide are obtained. Based on K-means clustering method, the image is segmented by K-means clustering, the edge of the image is extracted, and the landslide edge is obtained by region growth method. Finally, the area caused by landslide disaster is calculated. In this paper, the corresponding program is written for the proposed algorithm, including image smoothing, image segmentation, edge processing, image filling and so on. The function of landslide boundary extraction and landslide disaster area estimation is well realized.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P642.22;P237

【参考文献】