提出一种快速高精度Fourier-Mellin变换图像配准方法,可提高高光谱成像过程中光谱复原的速度与精度。该方法运用两次局部上采样相位相关法,一次在对数极坐标域中估计旋转和缩放参数,另一次在笛卡尔坐标系中估计平移参数。引入梯度预处理过程,使算法更加稳健。介绍了Fourier-Mellin变换的原理,该变换可实现旋转和缩放参数的解耦合。在传统相位相关法的基础上介绍了局部上采样相位相关法。证明了梯度预处理过程的必要性。仿真结果表明,所提方法可以实现高精度的图像配准。

A fast and high precision imaging registration method is presented, which can be used to improve the speed and accuracy of spectral recovery in a hyperspectral imaging process. In this method, two local upsampling phase correlation methods are adopted. One is to estimate the rotation and scaling parameters in a log-polar space, and the other is to estimate the translation parameters in a Cartesian space. A gradient preprocessing process is introduced to make the method more robust. The principle of Fourier-Mellin transform is introduced and this transform can decouple the rotation and scaling parameters. Based on the traditional phase correlation method, the local upsampling phase correlation method is introduced. The necessity of the gradient preprocessing process is verified. The simulation results show that the proposed method can be used to achieve high precision image registration.