【作者】 刘莉；

【导师】 刘庆元；

【作者基本信息】 中南大学 ， 地理学， 2013， 硕士

【摘要】 在地理空间数据库中建筑物是核心地形要素之一,同时也是城市环境中不可缺少的重要组成部分以及人类活动的重要聚居地。而随着社会的发展,建筑用地不断的发生变化,加快空间数据库的更新也变得非常重要。遥感技术的发展,使得人类精确识别和提取地物信息,自动更新GIS数据库成为可能。因此,本文着重围绕如何从高分辨率遥感影像上提取建筑物轮廓展开研究,研究的主要内容如下：(1)利用影像中波段数值运算的原理与绿波段中植被的反射率比较大的特性,采用绿波段与红波段的差值运算方法,消减其他地物的亮度值,增加植被亮度值,经过形态学修复之后,采用简单灰度阈值分割的算法提取出植被,再进行掩膜处理,消除植被对后续建筑物提取的干扰；最后,采用相同分辨率的不同区域的遥感影像数据验证该植被提取方法的通用性。(2)详细分析多尺度分割算法原理和分割尺度的确定方法。采用多尺度分割算法和K邻近的分类方法将去除植被的影像进行分割和分类识别,得出建筑物轮廓；最后设计了三个个对比试验：1)仅仅利用灰度特征提取建筑物轮廓的结果与本研究采用方法的结果相比较,分析提取精度。2)本研究提取方法的结果与传统的基于像元的分类提取结果进行对比,结果表明：利用面向对象的思想对本研究中的高分辨率遥感影像进行分类的方法具有明显的优越性,提取精度明显提高。3)在相同分割条件下,对比分析采用知识规则分类提取结果与采用本研究方法提取结果,分析提取精度。(3)文章提出了一种基于数字化的方法对于建筑物提取的结果中漏分、错分的现象进行后处理,即使用3种不同的数字化方法(手扶跟踪数字化,自动跟踪数字化,GIS数据叠加匹配)分别对轮廓破坏建筑物轮廓进行了修复处理。为了加快提取速度,本文提出将提取的建筑物轮廓与GIS数据库中有的相同地区矢量数据匹配叠加,快速自动修复破坏了的建筑物轮廓。三种方法都是通过c#和arcengien平台开发实现。更多还原

【Abstract】 Building areas is one of the core terrain features in geodatabase, meanwhile it’s also an important and indispensable part of urban environment and settlements of human activities. With the social development, building land constantly changes speedly, and spatial database update also become very important. The development of remote sensing technology, makes identification and extraction the feature information from remote sensing for human to automatically update the GIS database possible. Therefore, this paper focus on how to extracted building outline from high-resolution remote sensing images, The main research work is as follows:(1)An algorithm based on the principle of the numerical computation among the bands and vegetation reflectance characteristics of green band which can Weaken the luminance value of the other feature, and increase the vegetation luminance value.Then After the the morphological repairing, we use simple grayscale threshold segmentation algorithm to extract vegetation outline, and with the Mask processing to eliminate the interference of vegetation to the follow-up extracting of building. Finally, we take the different regions of the same resolution remote sensing image data to verify the versatility of the vegetation extraction method.(2) First, a detailed analysis of multi-scale segmentation algorithm and method of scale determination is made, meanwhile, the vegetation is removed by the image segmentation algorithm. Then, the divided image is classified using K neighboring algorithm to obtain the outline of buildings; moreover, extraction results of multi-scale segmentation and classification results based on pixel are compared and the accuracy is analyzed. Finally, outline of buildings are obtain using the algorithm of edge detection and vectorization.(3)An accurate building extraction method is proposed based on digitization, because of the false and miss alarms in the method. The severely and general damaged building outline are repaired using three digitization methods(hand tracking digitization, the automatic tracking digital, vector data matching repair).Finally, the building outline and the same vector data in GIS database are overlaid to speed up the repair rate. The three methods are developed using c#and Arcgis Engine.更多还原