Comparison of unconventional inter-filtering methods to penalised-likelihood for space-invariant tomographs

Images reconstructed with iterative algorithms with conventional inter-iteration filtering as a method of regularisation have spatially varying resolution, which is also object dependent. Three different methods were previously proposed which overcome that problem. Here we compare these algorithms with inter-iteration filtering to penalised likelihood algorithms and in particular penalised likelihood with certainty based penalty (PLCB). We derive an analytic expression that relates the penalty term to the filter coefficients. The linearised local impulse response (LLIR) as well as noise properties present in these algorithms are nearly identical if this matching is performed. This is confirmed with simulations.