基于人工免疫算法的光学影像与SAR影像配准方法

A Novel Image Registration Method for Optical and SAR Satellite Images Based on Artificial Immunity Algorithm

摘要:

提出了一种基于人工免疫算法的光学影像和SAR影像配准方法，该方法从影像上的面状地物入手，仅从识别性较好的光学影像上提取面状地物，先随机给定一组配准参数，将光学影像上面状地物的坐标经仿射变换获得新的坐标，以转换后新坐标在SAR影像上对应区域的均质性为评价标准，并利用人工免疫算法对配准参数进行优化，从而得到影像配准结果.最后，利用WorldView 2和RadarSat 2影像的配准实验验证该方法的有效性，结果表明该方法配准精度可优于2像素.

Abstract:

A novel registration method for optical and Synthetic Apperture Radar(SAR) satellite images based on artificial immunity algorithm (AIA) was proposed. By taking into account of the sufficient spectral information from optical satellite image, areal features were extracted from optical image based on the image classification strategy. Then the affine transformation with initial transformation parameters was adopted to obtain the new coordinates of features according to the location of features in optical image. Finally, the transformation parameters were optimized based on AIA by using the homogeneity of corresponding areas in SAR image, which were determined by the new coordinates of areal features, as the criterion of the transformation parameters. WorldView 2 and RadarSat 2 images were used to verify the effect of the proposed image registration method for optical and SAR satellite images. It is proved that the RMSE of image registration is less than 2 pixels.