Image Denoising | Inverse Problem | TV prior | Dual objective | GP Algorithm | Proximal | python
Denoising with (Isotropic) TV Regularizer with Dual objective function with the Gradient Projection (GP) Proximal Algorithm - a python implementation
Use Isotropic (L2) TV prior for Regularization
Since the penalty functional is not smooth in Primal, solve the problem Dual space, where the objective function is convex
Implement the GP algorithm
Add (uniform) noise with different noise variances to the image
Vary the regularization hyperparameter
Observe how the objective is minimized over iterations (negative of the maximizing objective in dual was used to turn the problem to a minimization problem)
#imageprocessing #imageprocessingpython #python #algorithm #computervision #machinelearning #optimization