I am interested in reovering images affected by blur of known orientation and known span. Camera movement during capture is very small, blur span is of about 0-4px.

What is the most accurate way to model the blur PSF? For example, how would you model the blur PSF in these cases?

  • blur span: 0.6 pixels, blur orientation: 0°
  • blur span: 2 pixels, blur orientation: 0°
  • blur span: 3 pixels, blur orientation: 25°

I am actually interested in recovering as accurately as possible the details of the images down to single pixel resolution, so blur PSFs such as [.5 .5] and [.25 .5 .25] are significally different for my pourpose.

  • $\begingroup$ How to generate PSF is a question I am struggling with as well. In my system, I have a motion that I know the details (car is moving at constant speed) I need to eliminate motion blur, I couldn't find a way to make PSF function. $\endgroup$
    – Ktuncer
    Sep 25, 2012 at 13:43

1 Answer 1


Given that you're able to generate the blurring PSF in a matrix form you may use the approach in How to Solve Blind Image Deblurring with Total Variation (TV) Prior Using ADMM.

Additional trick to solve the problem is to work in a multiscale approach. Namely first solve for small images then upsacle them.


Your Answer

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

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