Median based Priors for Non-Uniform Resolution Compensation in Tomographic Image Reconstruction

M. Amad and A. Tod-Pokropek (UK)


Penalized-Likelihood, image reconstruction, quadratic penalty, and median priors.


Penalized-Likelihood (PL) image reconstruction methods produce better quality images than normal analytical methods. However, these images exhibit non-uniform resolution properties. A number of prior functions have been used to recover this non-uniform resolution. Quadratic Priors (QPs) are preferred for penalization due to their simplicity and their resolution characteristics have been studied extensively. However, images reconstructed with QPs exhibit non-uniform resolution properties. Here, we propose two new priors based on QPs and median priors (MPs) and evaluate their resolution characteristics. The recovered resolution for the Combined Priors (CPs), a combination of QPs and MPs, can be set between QPs and MPs by selecting appropriate values for hyper parameters. For this modified MPs (ModMPs), almost uniform spatial resolution can be achieved.

Important Links:

Go Back