matlab - Fourier transformation -

i'm writing program in matlab related on image hashing. loading image , performing simple down-sampling not problem.here code

clear all; close all; clc; %%load image = im2single(imread('cameraman.tif'));  %%perform filtering , downsampling gausspyramid = vision.pyramid('pyramidlevel', 2);                  j = step(gausspyramid, i); %%preprocessed image %%get 2d fourier transform of input image y = fft2(i); %%fft of input image 

the algorithm next assumes 2d fourier transform ( y in case ) must in form y(f_x,f_y) f_x,f_y normalized spatial frequencies in range [0, 1].i'm not able transform output of fft2 function matlab required algorithm.

i looked paper in question (multimedia forensic analysis via intrinsic , extrinsic fingerprints ashwin swaminathan) , found

we take fourier transform on preprocessed image obtain i(fx, fy). fourier transform output converted polar co-ordinates arrive @ i′(ρ, θ)

by this, seem mean i(fx, fy) fourier transformation of image. they're trying use fourier-mellin transformation, different simple fft2 on image. based on information found in set of slides,

if input image expressed in terms of coordinates natural logarithms of original coordinates, magnitude of ft insensitive change in scale of original image (since mellin transform in original coordinates)

and in files on matlab central file exchange, have additional work mellin-fourier transform; particularly, taking magnitude (which appears missing step), converting coordinates log polar , taking second fft2. it's unclear me why log of log polar coordinates omitted paper in steps required. see an implementation image registration here example. note code old, , transformimage method appears not exist; log polar transform.