中文摘要：

由于影像信息提取过程蕴涵的诸多不确定性以及土地类别描述语境信息的含糊性影响，遥感数据的常规土地利用分类面临诸多困难与挑战。而模糊分类系统作为一种最为强大的软分类器，能处理、分析和表征遥感信息中传感器测量数据的不精确性、土地类别描述中的含糊性以及模型模拟中的不严密性，从而输出更能表达人类知识缺陷、更符合真实世界客观事实的分类结果，因此被认为是一种较好的土地利用遥感分类手段。本文以南京城市边缘带一样区为例，在采用地物导向分割技术对遥感影像分割的基础上，充分利用影像地物自身的光谱组合特征值以及其他空间形状、拓扑特征以及语境关系信息，按照模糊监督分类的过程来对研究区土地利用信息进行提取。研究结果表明基于遥感数据源的土地利用模糊分类系统可以获得比常规硬分类手段更为合理、信息含量更为丰富的输出结果。

英文摘要：

Application of remote sensing data in landuse classification often comes cross some difficulties and problems that originate from various types of uncertainty associated with image information extraction and ambiguity of the linguistic rules involved in the context information concerning dependency between features and landuse. Fuzzy classification system, as one of the most powerful soft classifiers, is capable of incorporating inaccurate sensor measurements, vague class descriptions and imprecise modeling in the analysis process, and outputting classification results that better demonstrate the limitation of human knowledge and the real world. Therefore, fuzzy classification is considered as a better method in landuse mapping based on remote sensing data. In this paper, a case study of the periurban Nanjing was carried out to extract landuse information by means of the supervised fuzzy classification, based on object-oriented segmentation and the resultant so-called image object information of not only spectral values, but also feature space using shape and topological features. Results indicte that fuzzy classification of landuse based on remote sensing data could achieve a more reasonable and meaningful result, in comparison with conventional rigid methods.