Image Deblurring and Super-resolution by Adaptive Sparse Domain Selection and Adaptive Regularization

Abstract

As a powerful statistical image modeling technique, sparse representation has been successfully used in various image restoration applications. The success of sparse representation owes to the development of l1-norm optimization techniques, and the fact that natural images are intrinsically sparse in some domain. The image restoration quality largely depends on whether the employed sparse domain can represent well the underlying image. Considering that the contents can vary significantly across different images or different patches in a single image, we propose to learn various sets of bases from a pre-collected dataset of example image patches, and then for a given patch to be processed, one set of bases are adaptively selected to characterize the local sparse domain. We further introduce two adaptive regularization terms into the sparse representation framework. First, a set of autoregressive (AR) models are learned from the dataset of example image patches. The best fitted AR models to a given patch are adaptively selected to regularize the image local structures. Second, the image non-local self-similarity is introduced as another regularization term. In addition, the sparsity regularization parameter is adaptively estimated for better image restoration performance. Extensive experiments on image deblurring and super-resolution validate that by using adaptive sparse domain selection and adaptive regularization, the proposed method achieves much better results than many state-of-the-art algorithms in terms of both PSNR and visual perception.

35 pages. This paper is under review in IEEE TIP

Cited by in corpus (29)

• A survey of sparse representation: algorithms and applications
• Denoising Prior Driven Deep Neural Network for Image Restoration
• CT Super-resolution GAN Constrained by the Identical, Residual, and Cycle Learning Ensemble(GAN-CIRCLE)
• Deep Edge Guided Recurrent Residual Learning for Image Super-Resolution
• Image Restoration Using Joint Statistical Modeling in Space-Transform Domain
• Image Compressive Sensing Recovery Using Adaptively Learned Sparsifying Basis via L0 Minimization
• Image Processing using Smooth Ordering of its Patches
• Working Locally Thinking Globally: Theoretical Guarantees for Convolutional Sparse Coding
• Random Walk Graph Laplacian based Smoothness Prior for Soft Decoding of JPEG Images
• Trainlets: Dictionary Learning in High Dimensions
• Learning Super-Resolution Jointly from External and Internal Examples
• Interpretable Detail-Fidelity Attention Network for Single Image Super-Resolution
• Fixed-Point and Objective Convergence of Plug-and-Play Algorithms
• A Sparse Representation Based Joint Demosaicing Method for Single-Chip Polarized Color Sensor
• Unified Single-Image and Video Super-Resolution via Denoising Algorithms
• Adaptive Transform Domain Image Super-resolution Via Orthogonally Regularized Deep Networks
• An Application of Generative Adversarial Networks for Super Resolution Medical Imaging
• Full waveform inversion with nonlocal similarity and model-derivative domain adaptive sparsity-promoting regularization
• Prompt-based Ingredient-Oriented All-in-One Image Restoration
• MPTV: Matching Pursuit Based Total Variation Minimization for Image Deconvolution
• FFT Multichannel Interpolation and Application to Image Super-resolution
• Extraction of Active Regions and Coronal Holes from EUV Images Using the Unsupervised Segmentation Method in the Bayesian Framework
• Blind Image Deblurring Using Row-Column Sparse Representations
• Estimation of the Regularization Parameter in Linear Discrete Ill-Posed Problems Using the Picard parameter
• Efficient Dictionary Learning with Sparseness-Enforcing Projections
• Across-scale Process Similarity based Interpolation for Image Super-Resolution
• Geometry-Aware Neighborhood Search for Learning Local Models for Image Reconstruction
• Weighted Encoding Based Image Interpolation With Nonlocal Linear Regression Model
• Subjective and Objective Quality Evaluation of Super-Resolution Enhanced Broadcast Images on a Novel SR-IQA Dataset