I succeeded to 2d FFT of image. But

You don’t need that line calling methods. Out of curiosity, what aspect do you find particularly complicated? Images tries not to guess your intentions; with a color image, fft(aa) would be ambiguous: do you want it of the grayscale-converted image like here, or perhaps you want the red color channel with fft(red.(aa)), or perhaps the u channel of an Luv-converted image with fft((c->Luv(c).u).(aa)). Or is it the fftshift that worries you?

5 Likes