基于超像素分割和多方法融合的SAR图像变化检测方法

针对基于像素的合成孔径雷达(Synthetic Aperture Radar,SAR)图像变化检测会造成虚警较高、结果破碎的问题,提出一种基于超像素分割和多方法融合的SAR图像变化检测方法。首先引入基于简单线性迭代聚类(Simple Linear Iterative Clustering,SLIC)的超像素分割方法,通过对主辅图像进行联合分割,得到符合实际地物边界的超像素分割结果;同时,利用3种基于像素的变化检测方法获取初始变化检测结果;接着,利用超像素分割结果和初始变化检测结果进行两个层次的众数投票,去除检测结果中由于噪声引起的虚警和连通域中的孔洞。选取两个时相的苏州Radarsat-2单极化SAR图像开展变化检测实验,实验结果表明该算法在保持较高检测率和有效边界的基础上,能够显著降低虚警。

Combining Super\|pixel Segmentation and Multiple Difference Maps for SAR Change Detection

The traditional pixel\|based change detection methods give high false alarm rate and broken areas.In order to solve this problem,we present a novel change detection method that combines a segmentation approach and three pixel\|based Difference Maps (DM).In this paper,the Simple Linear Iterative Clustering (SLIC) super\|pixel segmentation is introduced into SAR images segmentation,which can preserve edges between different land cover types and perform on two SAR images simultaneously.Meanwhile,three pixel\|based DMs are utilized to gain the initial change masks.Then,the majority voting is utilized for the fusion of segmentation result and initial change masks.Two Radarsat\|2 images of Suzhou,china,acquired on April 9,2009 and June 15,2010,are used for our experiment.The experimental results demonstrate that our method can reduce the false alarm rate effectively,as well as preserve a good change rate.Besides,the edge of changed objects are well preserved.