基于跨接约束的高分辨率SAR影像与光学影像配准

Registration of High-resolution SAR Image and Optics Image Based on the Bridging-mode Constraint

为了实现高分辨率SAR影像与光学影像之间自动/半自动配准,提出了一种新颖、稳健的匹配算法。算法首先利用仿射变换进行SAR影像和光学影像粗匹配,简化了整体算法的处理复杂度;然后利用影像边缘稳健性,使用边缘提取算子分别对SAR影像和光学影像进行边缘提取,为后续精匹配做好了数据准备;最后使用基于边缘纹理跨接约束进行影像之间精匹配,方法引入了邻域配准约束机制,很好的解决了经典匹配多峰值效应,提高了算法稳健性和实用性。以国内机载高分辨率SAR数据和SPOT-5 PAN数据为例进行算法验证,实验结果表明该算法能实现自动/半自动的高分辨率SAR和光学影像之间的像素级配准。

In order to realize the automatic or semi-automatic registration of SAR images with high resolution and optics images, in this paper, a novel and robust algorithm is proposed. The algorithm contains three main steps. Firstly, speckle filtering is employed on SAR image, and a coarse matching between SAR and optics images is implemented, which simplifies the complication of the whole process. In the second step, image edges are extracted from SAR and optics images, respectively, which treat as the input data for the subsequent process. Finally, a bridging- mode constraint method based on the edge texture is used to achieve a fine match. Due to the constraint and correlation between the processing local image region and its neighbor regions, the multi-value problem can be well resolved, which results to the algorithm＇s robustness and application. Domestic Airborne high-resolution SAR data and SPOT-5 PAN data are used for experiments, and the results show the methods proposed in this paper work well in automatic or semi-automatic registration of high-resolution SAR images and optics images, with pixels accuracy.