This code explore the PnP-ULA sensitivity to a denoiser shift and a measurement model shift. It is the code of the paper Plug-and-Play Posterior Sampling under Mismatched Measurement and Prior Models.
The folder pnp_ula_GMM_2D contains a notebook with the experiment for GMM in 2D. A parameter analysis in this toy example is also develop.
The folder pnp_ula_images contains the code use to run PnP-ULA on images (in gray-scale and in color). A specific README.md explain how to use the code in that folder. Various results of PnP-ULA on different images can also be find in the folder pnp_ula_images/results.
Here are some video presented samples during the stochastic process. Note how the quality increase during this process and how the Markov Chain continue to discover new possibilities.
 |
 |
 |
 |