3

I'm trying to determine the point-spread-function (PSF) of a certain microscope by a deconvolution process of a measurement of a test pattern (representing the blurred image) and a simulation of this measurement (representing the non-blurred image). The "images" them self are lines out of a microscopic image of the pattern (a height profile so to say), in other words, I'm working in 1D space. (I then want to use the results on 2D problems, just to mention that by the way.)

My question now: Is there any option to utilize the deconvolution functions already given by MATLAB (deconvlucy, deconvwnr, deconvreg...) to get to the PSF (their input demands a blurred image and a PSF normally) or another approach to get the PSF out of these two data sets? I already tried to do a fft of the "images", dividing them and then doing a ifft of the division (PSF = ifft((fft(measured))/(fft(sim)))) but the PSF of this ansatz does not yield the desired result when used in a later deconvolution.

Adriaan
  • 17,741
  • 7
  • 42
  • 75
JoshLo
  • 31
  • 2
  • Maybe have a look at this answer https://stackoverflow.com/questions/16541709/inverse-convolution-of-image - I think this is what you want... – Tom Wenseleers Mar 14 '18 at 09:56
  • Also check https://stackoverflow.com/questions/22362701/would-richardson-lucy-deconvolution-work-for-recovering-the-latent-kernel/34143964 – Tom Wenseleers Mar 14 '18 at 09:57
  • And https://stackoverflow.com/questions/19150516/2d-deconvolution-using-fft-in-matlab-problems – Tom Wenseleers Mar 14 '18 at 09:58

0 Answers0