Accelerated, Sparsity-Aware Generalizations of Classical Algorithms for TomoPIV
Abstract
We set up a generalized optimization framework for the classical Row Action Methods, otherwise customarily employed in the TomoPIV community. This scheme allows us to bypass known caveats of the algebraic programs. The so-enhance methods enable i) handling explicit constraints on the signal; ii) accelerating the rates of convergence. Comparative numerical simulations reveal superior performance of the latter with no inflation of the complexity.
Origin : Files produced by the author(s)