基于边缘与SURF算子的SAR与可见光图像配准方法

SAR and visible image registration method-based on edge and SURF algorithm

  • 摘要: 鉴于SAR(synthetic aperture radar)与可见光图像的成像机理存在很大差别,使得其同名特征的提取和配准十分困难,但在某些情况下,这两类图像的边缘存在一定的相关性。提出一种基于边缘与SURF(speed-up robust feature)算子的图像配准方法。通过适当预处理增强图像间的共性,采用综合性能比较好的Canny算子提取两幅图像共有的边缘特征,在边缘图像的基础上提取SURF特征;通过比值提纯法进行特征点粗匹配,RANSAC(random sample consensus)算法剔除误匹配点,计算仿射变换模型从而实现SAR与可见光图像的自动配准。实验结果表明:该算法的正确匹配率为100%,均方根误差为0.852个像素,配准精度达到亚像素水平。

     

    Abstract: Due to the prominent difference of the imaging mechanism between visible and synthetic aperture radar (SAR) images, it is very difficult to extract common features and register, but in some cases, the edges of these two kinds of images have some certain correlation. According to the above problem, an image registration algorithm based on edge and speed-up robust feature (SURF) is proposed. Firstly, the similarity of these two images is enhanced through appropriately preprocessing, the common edge features are extracted by Canny operator which has good performance, and the SURFs are extracted from the edges of the images; then the features matching is done by the ratio purification method, the random sample consensus (RANSAC) algorithm is applied to remove the false matching points, and the affine transformation model is calculated to realize image automatic registration of SAR and visible image. Experimental results demonstrate that the correct matching probability of the proposed algorithm is 100% and the root mean square error is 0.852 pixel, moreover, the registration accuracy can achieve sub-pixel level,which proves the validity of the algorithm.

     

/

返回文章
返回