基于非线性SIFT框架SAR图像匹配

SAR Image Matching Using Improved SIFT Framework Based on Rolling Guidance Filter and ROEWA-Powered Feature

针对基于尺度不变特征变换(SIFT)的图像匹配算法性能受到SAR图像中严重斑点噪声而性能降低的问题,提出了一种改进的非线性尺度构建的SIFT算法,主要改进在于:在尺度空间构建阶段,该算法通过将滚动引导滤波器嵌入到尺度空间构造的过程中来生成多尺度图像金字塔,在去除斑噪的同时并保持边缘的方面表现出了较其他尺度空间构建算法更好的效果;在特征检测阶段,提出了一种使用ROEWA算子和Harris-Laplace检测算子相结合的特征点检测算法,有效地抑制SAR图像中的虚假特征点,并准确地提取具有高位置精度和低误差率的不变特征点。3种不同类型的仿真和真实SAR图像对该算法进行了检验,并与其他2种基于SIFT的方法相比较,实验结果表明,该算法在匹配精度和内联点比率方面可以实现更好的性能。

Severe speckle noise and complicated local deformation in SAR image will affect the performance of common SIFT-based SAR image matching algorithms.To address this issue,this paper presents a new SIFT-based algorithm using a novel nonlinear multi-scale space construction strategy and a new local feature detector.For multi-scale space construction,the multi-scale images are generated by embedding a rolling guidance filter(RGF)into the procedure of scale space construction.The results show that this approach is more accurate in edge-preserving despeckling when compared with other strategies.For feature detection,an algorithm using Harris-Laplace together with ROEWA is proposed to effectively suppress the false feature points and accurately extract invariant feature points with high positional accuracy and low error rate in a SAR image.Experimental results using synthetic and real SAR images demonstrate that the proposed algorithm can achieve a better performance in terms of matching accuracy and inliers ratio compared with other SIFT-based methods.