Say I have a signal that is not x-indexed. That is, the x-axis of the signal is the distance traversed by car and the y-axis is the heading direction of the car at the corresponding distance.

Can I apply the Fourier Transform to this signal? If so, what is the physical meaning of this transformation? I believe that the horizontal axis is no longer frequency any more. What is it in this case?


1 Answer 1


Yes you can. The unit of the "frequency" axis after the transform will be $m^{-1}$, and is known as spatial frequency.

For example, if there is a strong peak in the Fourier transform at $3.10^{-4} m^{-1}$, it means that your original curve exhibits a strong pattern that repeats at a scale of every $3.3km$, and from that you could infer that maybe the signal was recorded from a vehicle doing laps at the Monaco Grand Prix. The harmonics of this spatial frequency would contain a "signature" of the shape of the circuit.

A practical application of this is handwriting recognition - looking at shapes in the Fourier domain yields representations invariant to scaling, rotations, or more robust to deformations than the original data.

  • $\begingroup$ OMG. The answer is so awesome and it really enlightens me! Thank you so so much! I am asking this because I wish to use FFT to find out the similarity between two such signals. Because the two signals are not aligned on x-axis, so I think FFT might help. Could you please further give some comments? Yes just as you said, ultimately I wish to infer the road shapes. $\endgroup$ Sep 19, 2013 at 7:58
  • $\begingroup$ More information: stackoverflow.com/questions/18887200/… I really appreciate your kind help! $\endgroup$ Sep 19, 2013 at 8:49

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.