# One question on the sampling theorem in Fourier transform

1. Sep 18, 2015

### Star_Sky

Hello everyone,

The question that I have may not be fully relevant to the title, but I thought that could be the best point to start the main question!
I'm working on 2-D data which are images. For some reason, I have converted my data to a 1-D vector, and then transformed them to the frequency domain using Fourier transform. My principal idea is that some features repeat every n pixels, say 100 pixels, in the image, where the total size of the vectorized form of the image is N. Therefore, as I know, the frequency I'm looking for would be n. However, I guess this idea is not true at all. Further, when I take Fourier transform of my data, using Matlab, there is not the frequency of 100 within the frequencies that Matlab yields. I'm really confused because of this and hope you can help me and tell me where the problem is and how to solve that.

Thank you.

2. Sep 18, 2015

### Staff: Mentor

Welcome to the PF.

Can you post some sample images that you are working with?

Why are you converting the 2-D images to 1-D data? That seems to be a pretty random thing to do, IMO...

3. Sep 18, 2015

### Star_Sky

Thank you for the reply.
The images I'm working on cover the buildings of urban areas; that is, applying very high resolution satellite imagery. Particularly, my goal is to identify the buildings located within n pixels. I want to use the Fourier transform to get their frequency, and then perform the inverse Fourier transform to achieve only the n-pixel wide buildings! Of course, before applying inverse Fourier transform, I first convert the vectorized image to its original 2-D format. This is all I'm going to do using Fourier transform.