I am facing the following problem: I have an image and a contrast sentivity function (CSF). The CSF is a function defined in the frequency domain and it is just a band-pass filter. The problem is that it is a 1D filter so I do not know how to apply it to the Fourier transform of the image.

I thought about creating a 2D filter based on the revolution around the zero frequency of the 1D filter but I do not know if it is mathematically correct or if it is the common procedure.

Thanks in advance!


I believe what you want to do is take your 1D filter column vector, call it $x$, and create a 2D filter with $xx^T$. The result of this outer product (column vector times row vector) is a matrix that you can use to do element-wise multiplication with the 2D Fourier transform of the image.

| improve this answer | |
  • $\begingroup$ I am sorry. I accepted your answer but I realized that it is not what I wanted. If you do the outer product you do not get the revolution of the 1D function but only in the quartes. Nevertheless, I give you up vote for the idea (EDIT: It seems I am not allowed :S). $\endgroup$ – DOMiguel Aug 5 '15 at 7:23
  • $\begingroup$ Perhaps I don't understand what you are trying to do... it sounds like you want to rotate a vector in a polar coordinate system? $\endgroup$ – CMDoolittle Aug 5 '15 at 15:36
  • $\begingroup$ That's it, but with outer product you do not get an exact rotation. For example, the outer product of the function $y=x^2$ defined over the region $[-1,1]$ is not a paraboloid of revolution. $\endgroup$ – DOMiguel Aug 5 '15 at 15:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.