Momentum-Net: Fast and convergent iterative neural network for inverse problems
arXiv:1907.11818 · doi:10.1109/TPAMI.2020.3012955
Abstract
Iterative neural networks (INN) are rapidly gaining attention for solving inverse problems in imaging, image processing, and computer vision. INNs combine regression NNs and an iterative model-based image reconstruction (MBIR) algorithm, often leading to both good generalization capability and outperforming reconstruction quality over existing MBIR optimization models. This paper proposes the first fast and convergent INN architecture, Momentum-Net, by generalizing a block-wise MBIR algorithm that uses momentum and majorizers with regression NNs. For fast MBIR, Momentum-Net uses momentum terms in extrapolation modules, and noniterative MBIR modules at each iteration by using majorizers, where each iteration of Momentum-Net consists of three core modules: image refining, extrapolation, and MBIR. Momentum-Net guarantees convergence to a fixed-point for general differentiable (non)convex MBIR functions (or data-fit terms) and convex feasible sets, under two asymptomatic conditions. To consider data-fit variations across training and testing samples, we also propose a regularization parameter selection scheme based on the "spectral spread" of majorization matrices. Numerical experiments for light-field photography using a focal stack and sparse-view computational tomography demonstrate that, given identical regression NN architectures, Momentum-Net significantly improves MBIR speed and accuracy over several existing INNs; it significantly improves reconstruction quality compared to a state-of-the-art MBIR method in each application.
arXiv comment28 pages, 13 figures, 3 algorithms, 4 tables, submitted revision to IEEE T-PAMI
References in corpus (15)
- Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising
- Deep Convolutional Neural Network for Inverse Problems in Imaging
- Low-Dose CT with a Residual Encoder-Decoder Convolutional Neural Network (RED-CNN)
- Trainable Nonlinear Reaction Diffusion: A Flexible Framework for Fast and Effective Image Restoration
- MoDL: Model Based Deep Learning Architecture for Inverse Problems
- Compressed Sensing MRI Reconstruction using a Generative Adversarial Network with a Cyclic Loss
- CNN-Based Projected Gradient Descent for Consistent Image Reconstruction
- Plug-and-Play Priors for Bright Field Electron Tomography and Sparse Interpolation
- Convolutional Dictionary Learning: Acceleration and Convergence
- Deep BCD-Net Using Identical Encoding-Decoding CNN Structures for Iterative Image Recovery
- Compressed Sensing and Parallel Acquisition
- Convolutional Analysis Operator Learning: Acceleration and Convergence
- Relaxed Linearized Algorithms for Faster X-Ray CT Image Reconstruction
- Convolutional Analysis Operator Learning: Dependence on Training Data
- Algorithmic Design of Majorizers for Large-Scale Inverse Problems