Insights into analysis operator learning: From patch-based sparse models to higher-order MRFs

Abstract

This paper addresses a new learning algorithm for the recently introduced co-sparse analysis model. First, we give new insights into the co-sparse analysis model by establishing connections to filter-based MRF models, such as the Field of Experts (FoE) model of Roth and Black. For training, we introduce a technique called bi-level optimization to learn the analysis operators. Compared to existing analysis operator learning approaches, our training procedure has the advantage that it is unconstrained with respect to the analysis operator. We investigate the effect of different aspects of the co-sparse analysis model and show that the sparsity promoting function (also called penalty function) is the most important factor in the model. In order to demonstrate the effectiveness of our training approach, we apply our trained models to various classical image restoration problems. Numerical experiments show that our trained models clearly outperform existing analysis operator learning approaches and are on par with state-of-the-art image denoising algorithms. Our approach develops a framework that is intuitive to understand and easy to implement.

13 pages, 10 figures, accepted to IEEE Image Processing

References in corpus (2)

• Revisiting loss-specific training of filter-based MRFs for image restoration
• Learning ${\ell }_1$-based analysis and synthesis sparsity priors using bi-level optimization

Cited by in corpus (32)

• Trainable Nonlinear Reaction Diffusion: A Flexible Framework for Fast and Effective Image Restoration
• SAR Image Despeckling by Deep Neural Networks: from a pre-trained model to an end-to-end training strategy
• ${\ell }^1$-Analysis Minimization and Generalized (Co-)Sparsity: When Does Recovery Succeed?
• The structure of optimal parameters for image restoration problems
• On learning optimized reaction diffusion processes for effective image restoration
• Biomedical Image Reconstruction: From the Foundations to Deep Neural Networks
• A higher-order MRF based variational model for multiplicative noise reduction
• Bilevel methods for image reconstruction
• Learning Filter Bank Sparsifying Transforms
• Truncated Back-propagation for Bilevel Optimization
• Learning Co-Sparse Analysis Operators with Separable Structures
• A bi-level view of inpainting - based image compression
• Supervised Learning of Sparsity-Promoting Regularizers for Denoising
• An abstract convergence framework with application to inertial inexact forward--backward methods
• An adaptively inexact first-order method for bilevel optimization with application to hyperparameter learning
• Dyadic partition-based training schemes for TV/TGV denoising
• An Optimal Control Approach to Early Stopping Variational Methods for Image Restoration
• Gradient Networks
• Controlled Learning of Pointwise Nonlinearities in Neural-Network-Like Architectures
• Cost Function Unrolling in Unsupervised Optical Flow
• Learning optimal orders of the underlying Euclidean norm in total variation image denoising
• Accelerating GMM-based patch priors for image restoration: Three ingredients for a 100$\times$ speed-up
• Separable Cosparse Analysis Operator Learning
• Image Denoising via Multi-scale Nonlinear Diffusion Models
• Semantic-Aware Depth Super-Resolution in Outdoor Scenes
• Regularization of Inverse Problems: Deep Equilibrium Models versus Bilevel Learning
• Bilevel Learning with Inexact Stochastic Gradients
• Product-of-Gaussian-Mixture Diffusion Models for Joint Nonlinear MRI Reconstruction
• One dimensional fractional order $TGV$: Gamma-convergence and bilevel training scheme
• Techniques for Gradient Based Bilevel Optimization with Nonsmooth Lower Level Problems
• The Weighted Ambrosio - Tortorelli Approximation Scheme
• Higher-order MRFs based image super resolution: why not MAP?