Fig. 3
From: Anti-noise variational sparse Bayesian estimation ghost imaging based on 3Level factor graph
Workflow of the proposed CSGI framework. (Phase 1) KSVD Dictionary Learning (blue), Input: training image patches \(\textbf{Y}\); Process: sparse coding and dictionary update (details in Appendix A.1); Output: dictionary \(\textbf{D}\), sparse codes \(\varvec{\alpha }\). (Phase 2) 3L-Bayesian Reconstruction (green), Input: bucket measurements \(\textbf{B}\) in Eq. (4); Process: initialize and update parameters (details in Appendix A.4-A.7); Output: \(\hat{\varvec{\alpha }}_{\text {est}}\), \(\textbf{W}_{\text {est}} = \textbf{R}(\textbf{D}\hat{\varvec{\alpha }}_{\text {est}})\).
