Fourier Mellin Image Registration

An implementation of Fourier Mellin Phase Correlation for Image Registration

Description

This is a tidied up version of Adam Wilmer's Fourier-Mellin transform for simple image rotation, scale and translation. No GUI is included and some code is cribbed directly from his implementation. I'm totally new to Matlab, so please excuse any coding faux-pas I have committed here.

• Required Products: Image Processing Toolbox
• MATLAB release: MATLAB 7.6 (R2008a)

[Kenneth Eaton]

Quite a racy sample pic, don't you think? =)

[Robinson Laundon]

You know what Kenneth? I've been playing with this image during testing for Fourier Mellin (and other things) and thanks to you, for the first time, I've noticed the pixels to which you are referring. Now I look back over all of the papers I've been reading that make use of Lena, I can see they are appropriately cropped (!). I can only offer apologies and if I can work out how to modify my submission, some edits for those of a sensitive disposition.

[Kenneth Eaton]

No sensitive disposition here, I just found it amusing =). I figured it was just a miscrop, but that you might want to know before anyone started complaining. If you log in and click to your submission page, there should be an option on the right (in the blue box) to update/modify your submission. I think you can just modify the picture and leave everything else.

K

[Pradyumna A]

Nice, it was useful in building my program.

[Chaos]

bombs with UINT16 images

??? Error using ==> times
Matrix dimensions must agree.

Error in ==> Register at 129
    IB = hipass_filter(size(I2, 1),size(I2,2)).*abs(FB);

??? Error using ==> times
Number of array dimensions must match for binary array op.

Error in ==> Register at 129
    IB = hipass_filter(size(I2, 1),size(I2,2)).*abs(FB);

??? Subscript indices must either be real positive integers or logicals.

Error in ==> interp2>nearest at 563
    F = arg3(ndx);

Error in ==> interp2 at 230
    zi = nearest(ExtrapVal,x,y,z,xi,yi);

Error in ==> Register>transformImage at 400
    r=interp2(A,xx,yy,'nearest');

Error in ==> Register at 131
    L1 = transformImage(IA, SizeX, SizeY, SizeX, SizeY, 'nearest',
    size(IA) / 2, 'valid');

[Robinson Laundon]

Chaos, the comments at the top of the code state that it only works with 8-bit greyscale images, unfortunately.

[Yaron Kauffmann]

The rotation and translation works great. How can i add also the scaling to this code ?

[lama riad]

what does 1 and 2 mean in size(I,1),size(I,2)??????