利用SAR-FAST角点检测的合成孔径雷达图像配准方法

SAR Image Registration Using SAR-FAST Corner Detection

合成雷达孔径图像配准作为变化检测和图像信息融合的基础,对多时相SAR图像的解译具有重要作用。该文提出一种基于SAR-FAST角点检测的图像配准方法。首先,选用迭代引导平滑算法抑制斑点噪声对角点检测的影响;然后,以检测点为圆心,选择合适的检测半径,在圆周上选取检测窗口,统计与检测点不相似的窗口数量,判断检测点是否为角点;最后,对候选角点进行分析,根据其强度分布特点进一步剔除误检点。实验结果表明,SAR-FAST可以检测到足够数量且稳定性和重复性好的角点,应用于图像配准,也能获得较好的配准效果。

As the basis of change detection and image fusion, SAR image registration plays an important role in the interpretation of multi-temporal SAR images. This paper presents a method of SAR image registration based on corner detection using SAR-FAST, which is a customized version of Features from Accelerated Segment Test （FAST） for processing SAR images. The proposed method firstly employs rolling guidance filter to suppress speckle noise. Secondly, the candidate corner point is determined by quantitative analysis of the dissimilarities of the detection windows on the extended circle and the center window. Finally, the error detections are removed by analyzing the intensity distribution properties of the candidate corners. The experimental results show that SAR-FAST can detect a sufficient number of corners with stability and high repeatability, and when applying to image registration, it also can get better registration results.