【作者】 黎薇；

【导师】 过仲阳；

【作者基本信息】 华东师范大学 ， 地图学与地理信息系统， 2007， 硕士

【摘要】 目前，随着遥感技术的迅速发展，特别是遥感影像处理水平的提高，遥感在社会各领域的应用日益广泛。在城市规划方面，利用遥感可以实现土地利用的动态监测，空气质量的监督控制，城市生态环境规划建设等。近年来，国内外许多城市将遥感技术应用到绿地信息提取中，以动态掌握绿地覆盖面积，优化绿地空间结构，这不仅可以实现城市绿地的整体规划，同时对于改善生态环境，提高城市的可持续发展潜能具有重要的实际意义。论文以上海市2003年的ETM遥感影像为数据源，对影像进行校正、裁剪、光谱增强等一系列的预处理，然后通过监督分类、非监督分类和模糊C均值聚类三种方法分别对上海市地物类型进行分类，提取绿地信息；并以普陀区为例，对三种分类方法进行了比较分析。研究结果表明，从分类总体效果来看，模糊C均值分类方法优于监督分类，而监督分类又优于非监督分类。然后，根据三种方法得到的上海市绿地信息，对上海市绿化建设现状进行了分析，在此基础上提出了未来的发展目标及建设措施。最后，提出了模糊C均值方法的尚存的一些不足之处。论文的特色及创新之处在于：针对遥感影像中的混合像元，引入了目前最流行的模糊C均值聚类方法，它是结合模糊集理论和K-均值聚类方法而提出来的适合进行软划分的模糊聚类分析方法。FCM方法对于遥感影像中存在的模糊性和不确定性，能较好的根据隶属度对混合像元加以判别，这对于反映地物的实际情况，能得到更加准确的效果。更多还原

【Abstract】 Nowadays, with the rapid development of remote sensing technology, especially the improvement of remote image processing, remote sensing has be applied in the various social fields more and more widely increasingly. In the way of city planning, remote sensing can be used in dynamic supervision of land utilizing, monitoring and management of atmosphere quality, programming and construction of city entironment and so on. In the recent years, many cities of domestic and overseas have applied remote sensing to green space information extraction, in order to find out area green cover dynamic and optimize the spatial structure of green space. This not only can do the holistic planning of green space, but also has the effective meaning for ameliorating the entironment benefit and increasing the city’s potential of continuable development.ETM RS image of shanghai of 2003 were taken as the data resources in this thesis. In the first place, do the emendation, subset, spectral enhancement and so on. Then clustering was done by unsupervised classification, supervised classification and Fuzzy C means, by which the areas of shanghai were classified and information of green space were extracted. Thirdly, take example for Putuo district, comparison and analysis of three class means were taken. The research result shows the whole effect of three mean are as follows: the FCM is the best of these three and the supervised classification is better than supervised classification.After that, analyzing of actuality of shanghai green structure was done, and development goal and construct measure were put forward on the accordance of analysis result. Finally, the shortcomings of FCM were analyzed at the end of the thesis.The characteristic and innovation of this thesis is that: Fuzzy C means, which is one of the most popular classified method at recent, is a fuzzy method suitable for soft clustering on the accordance of fuzzy set and K-means clustering. FCM is good for mixed pixel because it can compartmentalize by degree of membership. Using the method can gain the exacter result based on the factual condition of land.更多还原