基于NDR-CVA的遥感影像自适应变化检测

Adaptive change detection of remote sensing images based on NDR- CVA

针对遥感影像自适应变化检测中差异影像的构造和阈值分割两大问题,本文首先提出了基于邻域信息的变化向量差值比值融合法(NDR-CVA),实现了差值比值融合法的扩展,有效提取了变化信息;然后,利用改进的二维最大类间方差法(2D-OTSU)进行阈值分割,提高了变化检测过程的自动化程度;最后,利用Landsat 7和Landsat 5遥感影像进行实验对比,并通过精度评价,证明了NDR-CVA算法是一种有效的遥感影像自适应变化检测方法。

For the two key issues of difference image constructing and threshold segmentation in remote sensing image adaptive change detection,a fusion method of change vector analysis based neighborhood difference and ratio is introduced firstly in this paper to realize the wide use of the fusion method in remote sensing image,and effectively extract the change information.Then,the improved fast 2D-OTSU algorithm is used to segment the threshold,and to improve the automation of change detection process.Finally,it is proved that the method is an effective way in remote sensing image adaptive change detection through the experimental comparison and precision evaluation by using the Landsat 7 and Landsat 5 remote sensing images.