SPATIAL RESOLUTION PROPERTIES OF PENALIZED WEIGHTED LEAST-SQUARES TOMOGRAPHIC IMAGE RECONSTRUCTION WITH MODEL MISMATCH

This paper presents an analysis of the spatial resolution properties of tomographic image reconstruction based on a regularized least-squares objective function. The derivations are based on an idealized space-invariant tomographic system having a continuum of radial and angular samples. The analysis accounts for mismatch between the true radial point-spread function (PSF) of the measurements and the PSF used in the system model for image reconstruction. An angular-dependent weighting is also included, which provides insight into the asymmetric PSFs observed in images reconstructed by penalized weighted least-squares methods.