I currently researching the phase (between real and imaginary part) of the fourier transformation. It seems that one can use it for detection.

For the signal processing I use two fft transforms. One to compress range and one to compress azimuth. Then I want to apply detection. At this point calibration and corrections are done. The scene is stationary anyway thus one can ignore doppler correction.

Claim: For a stationary target the phase of the cross-fft is stable and I can do the following:

$\text{Var}(\angle_N(Re(X_{i,j}), Im(X_{i,j})) = \sigma_{i,j}$,

whereas $N$ the number of used samples and $X$ is the spectrum of the signal (after cross fft) with dimension ${i,j}$ for range and azimuth.

For a phase stable pixel of the data matrix the variance $\sigma$ should be low. To find the most stable pixels I can do:

$f(\sigma) = \text{sort}(\sigma_{i,j})< T = \text{detections}$,

whereas $T$ can be constant or adaptive.

The method is intuitive and easy to implement. It cost alot ressources, but this point can be excluded at this point.

I cannot find any paper where someone researched ground based radars that image a stationary scene (forest, house, whatever) without using SAR. Also I cannot find a paper that exploit the phase of the fourier transformation.

Has anyone a clue if this is worth to investigate further or do I missunderstand a common pitfall which is maybe too obvious for everyone else? I am grateful for papers, references, ideas, corrections, ...

This is non-profit academic research (thesis)

  • $\begingroup$ How do you compute the variance of a single sample? $\endgroup$ Commented Jul 12, 2023 at 20:26
  • $\begingroup$ I dont, I store N frames of the recorded scene and calculate the variance with 12-25 of them. $\endgroup$
    – magi
    Commented Jul 18, 2023 at 15:35
  • $\begingroup$ It's not clear to me what you are proposing, and I am going to guess from the lack of any other responses that it is not clear to others either. The idea of a phase-only detector seems far-fetched and, honestly, like a bad idea. $\endgroup$ Commented Jul 18, 2023 at 18:46


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.