Autocorrelation matrix with FFT2

Hi guys, i have a problem with fft2.
I must find the autocorrelation matrix of a matrix s = 4x10000.

The first and right solution is R = s*s'; <--matrix 4x4

Now i want obtain the same result with fft2,so:

o = fft2(s);
kl = o*o';
R = ifft2(kl);  <- matrix 4x4

But i see that the fourth column is different, while the others yes..Why??

On Sunday, October 20, 2013 9:37:33 PM UTC+13, Ste_ee wrote:
> Hi guys, i have a problem with fft2.
> 
> I must find the autocorrelation matrix of a matrix s = 4x10000.
> 
> 
> 
> The first and right solution is R = s*s'; <--matrix 4x4
> 
> 
> 
> Now i want obtain the same result with fft2,so:
> 
> 
> 
> o = fft2(s);
> 
> kl = o*o';
> 
> R = ifft2(kl);  <- matrix 4x4
> 
> 
> 
> But i see that the fourth column is different, while the others yes..Why??

Over to you Vlad. Vlad the inhaler will help you out.

>Hi guys, i have a problem with fft2.
>I must find the autocorrelation matrix of a matrix s = 4x10000.
>
>The first and right solution is R = s*s'; <--matrix 4x4
>
>Now i want obtain the same result with fft2,so:
>
>o = fft2(s);
>kl = o*o';
>R = ifft2(kl);  <- matrix 4x4
>
>But i see that the fourth column is different, while the others
yes..Why??
>
>

Try asking at dsp.stackexchange.com, maybe provide an example with some toy
input where the expected output is known....	 

_____________________________		
Posted through www.DSPRelated.com

On Sun, 20 Oct 2013 22:41:53 -0700 (PDT), gyansorova@gmail.com wrote:


>
>Over to you Vlad. Vlad the inhaler will help you out.

Hah, that's a good one.

Mark