lamar_air

If you are familiar with FFT's

I have a Fast Fourier Transform component which receives a 1D array of real numbers and a 1D array of imaginary numbers. 2 arrays will be produced (one for the real values and one for the imaginary values) when the imput is converted from the time domain to the frequency domain.

I need some more explanation on how to do this part:
To make a 2D FFT you can use the 1D fft (described above) by performing the 1D FFT on all rows then all columns of the 2D input. This will give a 2D output.

Thanks

Bolek

You cannot do it this way because if you first transform all rows and then transform the columns, the second transform will run over the results of the first transform, not the original signal. 2D Fourier transform operates over the original signal in both dimensions.

lamar_air

If you do an FFT on each row first then the results of the first will be now in the matrix. Then you could do an FFT on each column which you are right won't use the origional values but you've already used all of the origional values once.

TeraBlight

To me, this sounds like the input and output arrays are separate, which means that what Bolek said wouldn't be an issue either way - the original data will be available throughout, if you want it to.
However, first google hit for "2D FFT" tells us that

"2 dimensional Fourier transforms simply involve a number of 1 dimensional fourier transforms. More precisely, a 2 dimensional transform is achieved by first transforming each row, replacing each row with its transform and then transforming each column, replacing each column with its transform. Thus a 2D transform of a 1K by 1K image requires 2K 1D transforms. This follows directly from the definition of the fourier transform of a continuous variable or the discrete fourier transform of a discrete system."

Sound good?

lamar_air

Sounds like the confirmation i was looking for. Merci