基于稀疏快速傅里叶的互相关图像配准

Cross-Correlation Image Registration Based on Sparse Fast Fourier

图像配准是影像处理领域的重要应用之一.其中基于傅里叶梅林变换的图像配准方法具有精度高,对明暗变化、部分遮挡、噪声影响等具有很好的鲁棒性等优点而被广泛应用.在此基础上提出了一种更有效率的互相关算法即稀疏快速傅里叶梅林算法,可以使时间复杂度与像素个数N呈亚线性关系,因此算法可以呈指数倍的效率提升,且不影响配准效果.

Image registration is one of the important applications in the field of image processing. The image registration method of Fourier Merlin transform which is widely used has the advantages of high precision and good robustness against change in light and dark, partial blocking, noise influence and so on. Therefore a more efficient cross correlation algorithm, sparse fast Fourier Merlin algorithm, is proposed in the paper. Because this method can make the relationship between the time complexity and the number of pixels N to be sub-linear, the algorithm can be exponentially higher efficiency and does not affect the registration effect.