论文部分内容阅读
合成孔径雷达(SAR)的成像过程使其高分辨图像的几何形变呈现局部性。针对高分辨SAR图像精确配准问题,本文提出一种基于邻域重构模型的局部转换函数.邻域重构模型首先采用重构系数刻画参考图像中每个像素点的几何位置;接着给出了一种重构控制点的选择标准使每个像素的配准误差达到最小;最后根据重构系数及控制点坐标对输入图像进行再抽样以实现配准。与经典分片线性映射相比,该模型从理论上给出了一种区域剖分准则:对于每个像素选取能使配准误差能达到最小的控制点作为重构控制点。对模拟数据和真实SAR图像进行了试验,结果表明,该模型能有效地提高配准精度。
The imaging process of Synthetic Aperture Radar (SAR) renders the geometry of high-resolution images locally distorted. For the accurate registration of high resolution SAR images, this paper proposes a local transformation function based on the neighborhood reconstruction model, which first uses the reconstruction coefficients to characterize the geometric location of each pixel in the reference image. A selection criterion for reconstructing control points minimizes the registration error of each pixel. Finally, the input image is resampled according to the reconstructed coefficients and control point coordinates to achieve registration. Compared with the classical patch linear mapping, this model theoretically provides a region partitioning criterion: for each pixel, the control point which can minimize the registration error is selected as the reconstruction control point. The simulation data and the real SAR image are tested. The results show that the model can effectively improve the registration accuracy.