why ripples are forming after IFFT(theoretical explanation needed)?

I have an intensity image. I want to know how these ripples are formed after IFFT theoretically? And how it is dependent on orientation from the origin and widths?

[Image Analyst]

Why did you take down the image????

Answer 1

[Image Analyst]

The ripples could be formed if you did a low pass filter. It's the high frequencies that help form flat regions, steps, edges, sharp ridges, etc. -- they keep the surfaces flat and straight. If you remove those with a low pass filter, all you'll be left with are the low frequencies with no high frequencies to keep the surfaces and profiles straight and flat.

[kumara dommeti]

I do not want to remove. I want to know how these are formed and its impact if I change orentation.(x(10:20,10:20)=randn(11,11);) Can you explain me theoretically how these are forming?

[Image Analyst]

There are no correlated ripples in your random noise. For that line of code, each pixel is independent of its neighbor. It's white noise and the fft would theoretically be a flat value since there would be spatial frequencies at all values. Actually since you have a finite image (your image is multiplied by a 2-D rect function), your fft would be a flat plane convolved with a 2-D Sinc function.

And I told you how you'd get ripples. If you have for example square ridges and you remove the high frequencies in the spatial frequency domain, the square ridges will now have ripples on them and be blurry.

[kumara dommeti]

What is the impact of placing x(10:20,10:20)=randn(11,11) near and far from the origin?

[Image Analyst]

Well of course the image would look different. And so would the fft, especially the phase image.