constructing a 1D using ifft

Hi there,

I get confused with all the website explaining how to construct an image from k-space using inverse fourier transform. I understand I need use abs(ifftshift(ifft(sig))) in order to produce a 2D image. Can someone please tell what I need to use in order to get a 1D image instead?

Best wishes

AA

[Matt J]

I assume you really meant

abs(ifftshift(ifft2(sig)))

[Azza Ahmed]

I guess you are right. But I found some people constructing a 2D using : abs(ifftshift(ifft(fftshift(sig)))) not abs(ifftshift(ifft2(sig))) and this where I am confused.

Answer 1

[Matt J]

As long as sig is 1D, you should be getting 1D output.

    >> sig=1:5;  %1D signal

    >> abs(ifftshift(ifft(sig))) %1D output

    ans =

        0.5257    0.5257    0.8507    3.0000    0.8507

Answer 2

[Image Analyst]

What is k-space? Do you mean "frequency space," or the "Fourier domain"? And your code will give a 1D "image" - it does not produce a 2D image. If you are dealing with 2D images, you would use the "2" version, such as fft2() and ifft2(). Why do you think you're getting a 2D image when you use those 1D functions like you showed? What does "whos" show for your output?

[Azza Ahmed]

Yes, indeed. I am constructing a 1D from my output. But why do people sometime use abs(ifftshift(ifft(fftshift(sig)))) to construct a 2D?

Highly appreciate your help with this

AA

[Image Analyst]

sig must have already been run through fftshift(). And you don't get a 2D image from that so you'd have to do it for every line and every column of your image - that essentially what fft2() does. Don't bother with that unnecessary complicated way of doing it, just use fft2 and make it easy.