基于光谱和空域信息的城区变化检测方法研究

目前的变化检测研究主要集中在利用中、小比例尺遥感图像进行自然环境等方面的变化检测,如草场的季节性变化、灾害检测、植被分布变化、土地使用规划等。然而随着城市的快速发展,为了满足城区管理规划中对道路、房屋等变化细节进行分析的需求,并由于高分辨率卫星诸如Ikonos,Quickbird等的出现,我们有必要并且有可能开发出一套实用、有效的、可靠的城区自动变化检测系统。针对城市区域的复杂性及其高分辨率卫星影像的配准误差问题,本文提出一种通过模糊逻辑结合光谱特征和空域特征的城区变化检测方法,以期望减小图像整体及局部配准误差对变化检测精度的影响。

An urban change detection method using spectral and spatial features

The current researches on change detection are mostly focused on environmental scope using low spatial resolution remotely sensed imagery, such as seasonal change detection of pasture, change detection of vegetation distribution, disaster detection, land use regularization, etc. However, along with the rapid development of cities, detailed analyses on roads' and buildings' changes are needed for city planning, and with the advent of high spatial resolution satellites such as IKONOS and QUICKBIRD, it is necessary and possible to develop a set of effective, helpful, reliable unsupervised urban change detection system. Aiming at the registration problem aggravated by the complexity of urban area, this paper introduce an urban change detection method, which is robust against registration error using spectral features and spatial features combined by fuzzy logic.