Pseudo-log-polar Fourier transform for image registration

A new registration algorithm based on pseudo-log-polar Fourier transform (PLPFT) for estimating large translations, rotations, and scalings in images is developed. The PLPFT, which is calculated at points distributed at nonlinear increased concentric squares, approximates log-polar Fourier representations of images accurately. In addition, it can be calculated quickly by utilizing the Fourier separability property and the fractional fast Fourier transform. Using the log-polar Fourier representations and cross-power spectrum method, we can estimate the rotations and scalings in images and obtain translations later. Experimental results have verified the robustness and high accuracy of this algorithm.