多波段遥感影像变化检测中差异影像构造研究

Research on constructing difference image in change detection of multiband remote sensed image

针对于多波段遥感影像变化检测中构造差异影像和确定变化阈值这2个关键问题展开研究,提出了融合法与主成分分析法结合构造差异影像的方法,实现了融合法对于多波段遥感影像的应用扩展,有效地抑制了噪声并实现了变化信息的有效集中和准确提取,再利用结合直方图的EM算法确定变化阈值,提高了流程的自动化程度,最终完成变化检测。最后利用实际遥感影像进行了实验,并通过与差值法和比值法的对比,证明了这种方法可以显著地提升变化检测的正确检测率,是一种有效的多波段遥感影像变化检测方法。

In this paper, research is aimed at two key issues of constructing difference image and obtaining of threshold in change in multi-band remote sensed image change detection. Fusion method and PCA method is firstly used to construct difference image, and realizes the expand use of Fusion method in multiband remote sensed image, and effective suppress the noise and accurately extract the change information. Then,the EM algorithm based on histogram is used to obtain change threshold and complete change detection. At last, through the experimental comparison with the traditional difference method and ratio method proves that this method is an effective way to enhance the correct detection rate of change detection in multiband remote sensed image.