一种基于快速傅里叶变换的分区域图像配准方法

为提高图像配准的速度,提出了一种基于快速傅里叶变换的分区域图像配准方法.在图像中选取一块小区域代替整体区域,根据对数-极坐标变换、傅里叶变换的特性确定图像间的比例和旋转变化,利用傅里叶的相位相关技术确定图像间的平移关系,从而求出图像的配准参数,最后利用这个配准参数对待配准图像进行全局配准.此方法改进了利用全图求取配准参数的过程,利用小区域变换得到配准参数,减少了运算量,提高了运算速度.

An Image Registration Method Based on Subregional Fast Fourier Transformation

In order to improve the speed and precision of image registration, a method of image registration based on subre-gional fast Fourier transformation was proposed. A small area was selected in the image instead of the whole area, and then the ratio between the images and the rotation change was got based on the property of log-polar coordinate transformation and Fourier transformation, and the phase correlation technique （energy spectra） was used to determine the image transla-tion. Using those registration parameters, the global registration can be done. The method can improve the processing speed because of using a small area instead the whole image.