基于Brovey融合的卫星影像土地覆盖变化检测

LAND COVER CHANGES DETECTION BASED ON BROVEY FUSION TECHNIQUE OF SATELLITE IMAGE

随着高空间分辨率卫星影像的发展,如ETM+,QuickBird,IKONOSE等高空间分辨率卫星,近20年来已相继开发了许多影像融合的方法,如HIS,PC,HPF,SFIM,SVR,Wavelet和Brovey等融合法.传统的影像融合是将时相相同或相近的影像进行融合,以提高影像的空间分辨率和光谱分辨率,提高影像的目视可解度和分类精度等.而本文却是对不同时相的影像进行融合,用Brovey融合法将不同时相的TM(1986年7月26日)和ETM+(2000年5月4日)的Pan波段影像进行融合,再对融合后的影像采用无监督分类和PC2分析两种方法将两时相的土地覆盖变化区域提取出来.研究证明融合法能快速、简便、准确的检测出土地覆盖变化的区域.

There are many image fusion methods developed in recent years with the development of multi-sensor, multi-temporal and multi-spectral remote sensing technologies. Of these, the HIS, PC, HPF, SFIM, SVR, Wavelet and Brovey fusion methods are widely used. However, the traditional image fusion is a technology that usually merges the images with the same temporal or near temporal data to increase the image spatial resolution and multi-spectral resolution. This paper is to merge different time serial TM/ETM+ images. After this process, we can apply the unsupervised classification method and the second principal component method on the fusion image to detect and extract land cover change areas. To evaluate accuracy of the methods, we use post-classification method to classify TM image of 1986 and ETM+ image of 2000, and then extract land cover change areas. After this, the detected results of unsupervised classification and the second component analysis were compared with the post-classification method. The comparison reveals that the two methods used in this study have much higher accuracy than the post-classification method. The study indicates that these two methods can quickly and efficiently detect land cover changes.