基于Fourier-Mellin变换的全景图像拼接与曲线匹配

基于Fourier—Mellin变换的图像配准方法，本文对于图像拼接与曲线匹配进行新探讨。该方法在不需要准确控制相机运动，不需要知道相机的焦距等内部参数及检测图像特征，在配准精度要求不是很高的情况下，直接生成的全景图像可以满足很多实际需求；先将图像曲线转化为二值图像，然后应用Fourier—Mellin变换对这些二值图像进行配准，从而达到两条曲线的匹配。

An Image Mosaics And Curve Matching Based on Fourier - Mellin Transform

Based on the Fourier - Mellin transform and the image registration, The image mosaics with curve matching has been restudied In this paper. The method adopted here does not need accurate control of the camera motion and to know the focal length of the internal parameters and image feature detection . The registration requirement is not very high accuracy in the cases, the direct generation of panoramic images can satisfy many practical needs . But firstly the images curve need to be turned into two binary images, then the Fourier - Mellin transform of these two values image registration are used to match the two curves.