Abstract:The importance of regularization has been well established in image reconstruction -- which is the computational inversion of imaging forward model -- with applications including deconvolution for microscopy, tomographic reconstruction, magnetic resonance imaging, and so on. Originally, the primary role of the regularization was to stabilize the computational inversion of the imaging forward model against noise. However, a recent framework pioneered by Donoho and others, known as compressive sensing, brought the role of regularization beyond the stabilization of inversion. It established a possibility that regularization can recover full images from highly undersampled measurements. However, it was observed that the quality of reconstruction yielded by compressive sensing methods falls abruptly when the under-sampling and/or measurement noise goes beyond a certain threshold. Recently developed learning-based methods are believed to outperform the compressive sensing methods without a steep drop in the reconstruction quality under such imaging conditions. However, the need for training data limits their applicability. In this paper, we develop a regularization method that outperforms compressive sensing methods as well as selected learning-based methods, without any need for training data. The regularization is constructed as a spatially varying weighted sum of first- and canonical second-order derivatives, with the weights determined to be adaptive to the image structure; the weights are determined such that the attenuation of sharp image features -- which is inevitable with the use of any regularization -- is significantly reduced. We demonstrate the effectiveness of the proposed method by performing reconstruction on sparse Fourier samples simulated from a variety of MRI images.

Multi-Parameter Regularization Methods for High-Resolution Image Reconstruction with Displacement Errors

High Resolution Image Reconstruction Via Multi-parameter Regularization Method

Discrepancy Curves for Multi-Parameter Regularization

Image Super-Resolutton Reconstruction Based On Adaptive Interpolation Norm Regularization

Time multiscale regularization for nonlinear image registration

Multi-parameter Regularization and Its Numerical Realization

New Regularization Models for Image Denoising with A Spatially Dependent Regularization Parameter

Multi-Objective Sparse Reconstruction With Transfer Learning and Localized Regularization

Data-Driven Multi-scale Non-local Wavelet Frame Construction and Image Recovery

Multiscale hierarchical decomposition methods for ill-posed problems

Multiscale Analysis for Ill-Posed Problems with Semi-Discrete Tikhonov Regularization

Image Super-Resolution Reconstruction by Huber Regularization and Tailored Finite Point Method

Two New Efficient Iterative Regularization Methods for Image Restoration Problems

Bregmanized Nonlocal Regularization for Deconvolution and Sparse Reconstruction

Iteratively Refined Image Reconstruction with Learned Attentive Regularizers

Medical Image Reconstruction with Multi-Level Deep Learning Denoiser and Tight Frame Regularization

Structurally Adaptive Multi-Derivative Regularization for Image Recovery from Sparse Fourier Samples

A Multi-Frame Image Super-Resolution Method.

Multi-Image Restoration Method Combined with Total Generalized Variation and L P -Norm Regularizations

A multi-parameter regularization model for image restoration

A New Highly Anti-Interference Regularization Method for Ill-Posed Problems