中文摘要：

提出了一种基于特征分解的无监督分割方法,对高分辨率遥感影像中的居民地进行提取。该方法通过小波分解多尺度特征,利用居民地内、外部结构差异以及平均光谱辐射强度差异构成特征空间,采用约束均值漂移算法进行特征空间自适应分解,实现居民地自动提取。实验结果表明,该方法能很好地消除高分辨率导致的影像高度细节化等因素对居民地提取的影响,有效提取居民地。

英文摘要：

This paper presents an unsupervised segmentation method based on the feature decomposition used for extracting residential areas in high resolution remote sensing image.The method was realized by multi-scale feature analysis with wavelet decomposition,constituting the feature space from differences of the internal,external structure in residential areas and the average spectral radiant intensity,making a self-adaptive segmentation by the constraint mean shift algorithm to texture features,and achieve the automatic extraction of residential areas.Experiment results show that the proposed method can eliminate the influence from over-detailed images and other factors caused by high resolution on the extraction of residential areas,and thus extract residential areas more effectively.