The structure of optimal parameters for image restoration problems

Abstract

We study the qualitative properties of optimal regularisation parameters in variational models for image restoration. The parameters are solutions of bilevel optimisation problems with the image restoration problem as constraint. A general type of regulariser is considered, which encompasses total variation (TV), total generalized variation (TGV) and infimal-convolution total variation (ICTV). We prove that under certain conditions on the given data optimal parameters derived by bilevel optimisation problems exist. A crucial point in the existence proof turns out to be the boundedness of the optimal parameters away from $0$ which we prove in this paper. The analysis is done on the original -- in image restoration typically non-smooth variational problem -- as well as on a smoothed approximation set in Hilbert space which is the one considered in numerical computations. For the smoothed bilevel problem we also prove that it $\Gamma$ converges to the original problem as the smoothing vanishes. All analysis is done in function spaces rather than on the discretised learning problem.

References in corpus (5)

• Insights into analysis operator learning: From patch-based sparse models to higher-order MRFs
• Revisiting loss-specific training of filter-based MRFs for image restoration
• Learning ${\ell }_1$-based analysis and synthesis sparsity priors using bi-level optimization
• The jump set under geometric regularisation. Part 2: Higher-order approaches
• Optimal Regularization Parameters for General-Form Tikhonov Regularization

Cited by in corpus (19)

• Solving ill-posed inverse problems using iterative deep neural networks
• Hyperparameter optimization with approximate gradient
• Bilevel parameter learning for higher-order total variation regularisation models
• A Bilevel Approach for Parameter Learning in Inverse Problems
• Analysis and optimisation of a variational model for mixed Gaussian and Salt & Pepper noise removal
• Dualization and Automatic Distributed Parameter Selection of Total Generalized Variation via Bilevel Optimization
• Learning nonlocal regularization operators
• Stability of saddle points via explicit coderivatives of pointwise subdifferentials
• Linearly convergent bilevel optimization with single-step inner methods
• Bilevel training schemes in imaging for total-variation-type functionals with convex integrands
• Dyadic partition-based training schemes for TV/TGV denoising
• Explorations on anisotropic regularisation of dynamic inverse problems by bilevel optimisation
• Learning optimal orders of the underlying Euclidean norm in total variation image denoising
• Real order (an)-isotropic total variation in image processing - Part II: Learning of optimal structures
• Regularization of Inverse Problems: Deep Equilibrium Models versus Bilevel Learning
• Infimal convolution of data discrepancies for mixed noise removal
• The Weighted Ambrosio - Tortorelli Approximation Scheme
• One dimensional fractional order $TGV$: Gamma-convergence and bilevel training scheme
• Techniques for Gradient Based Bilevel Optimization with Nonsmooth Lower Level Problems