基于粗配准SAR图像变化检测方法

Change Detection Method Based on Rough Registration SAR Images

合成孔径雷达(SAR)图像由于受相干斑噪声的影响,在对其进行配准及变化检测时,经典的适于光学图像的方法效果不佳。该文先通过选取少量控制点实现SAR图像的粗配准,然后利用直方图规定化对SAR图像进行图像灰度匹配处理,在此基础上根据待检测目标特性选取适当的阈值,利用滑动矩形窗实现改进的基于图像差值的变化检测。真实SAR图像的实验结果表明,该方法能有效克服图像照度差异和粗配准误差的影响,取得满意的检测结果。

Synthetic aperture radar （SAR） is affected by speckle noise seriously. It is difficult to use the classic means for optical images when we perform SAR images registration and change detection. In this paper a few control points（CPs ） are chosen to perform images registration firstly, and then histogram regulation is adopted to match luminance between images. At last, a proper threshold is selected and a glide rectangle window is utilized to perform change detection. Experiments on real SAR images demonstrate that this means can eliminate the image luminance difference and registration error and verifies its superiority.