In lab, I did a bunch of scans using a radiation source and a detector. My source emits a gaussian beam (whose dimensions I know), and my detector is a uniform 7mmx7mm square. These are stationary and I run my sample between them as the image below shows.

enter image description here

(sorry for not translating, "Provete" means sample, "Fonte" means source, and the grey thing is an arm that holds and moves my sample)

The thing is, the sample moves in steps way smaller than 7mm (usually 0.25mm) so my final image is convolved.

I've been trying to deconvolute my images with the Richardson-Lucy method by trial and error using different PSF's but with no success. I've tried gaussian, quadractic, a cone, truncated gaussian, pyramid, sinusoidal, a convolution between a square and a gaussian, and some others but to no success.

None of my results appeared any sharper than my original image. All of these I've tried through a range of dimensions and parameters, so in total I've tried thousands of possibilities.

To make sure I'm not committing some basic mistake I tried convolving the picture of "Lenna" and deconvolving it and it works beautifully.

Is there a better approach to this? Are there any obvious PSF's I'm not seeing?

  • $\begingroup$ Is the detector composed of a single Pixel? Is the beam from the source is constant all time (Namely as long as the detector move)? $\endgroup$ – Royi Apr 10 '20 at 22:54
  • $\begingroup$ Yes it is, and it absorbs radiation in a uniform way. And yes the source is always on, but it has a very fast response time (micro-seconds) so for practical effects it is as if the sample stops and I take a measurement at each point. $\endgroup$ – Frederico Costa Apr 10 '20 at 23:09
  • $\begingroup$ So basically the physical parameters to estimate is the step size in x and y of the arm which moves the detector. $\endgroup$ – Royi Apr 10 '20 at 23:17
  • $\begingroup$ I'm not sure I understand what you mean. The detector doesn't move, it is always centered in relation to the source. What moves is the sample, the object I'm taking an image of. And I know the steps in x and y, those where defined by me when I did the scans. What I can't find is a way to deconvolve this. $\endgroup$ – Frederico Costa Apr 10 '20 at 23:21
  • $\begingroup$ In that case the question isn’t clear. I don’t understand what is exactly captured in the detector. $\endgroup$ – Royi Apr 10 '20 at 23:23

Your Answer

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

Browse other questions tagged or ask your own question.