DR2-Net: Deep Residual Reconstruction Network for Image Compressive Sensing

Abstract

Most traditional algorithms for compressive sensing image reconstruction suffer from the intensive computation. Recently, deep learning-based reconstruction algorithms have been reported, which dramatically reduce the time complexity than iterative reconstruction algorithms. In this paper, we propose a novel \textbf{D}eep \textbf{R}esidual \textbf{R}econstruction Network (DR${}^2$-Net) to reconstruct the image from its Compressively Sensed (CS) measurement. The DR${}^2$-Net is proposed based on two observations: 1) linear mapping could reconstruct a high-quality preliminary image, and 2) residual learning could further improve the reconstruction quality. Accordingly, DR${}^2$-Net consists of two components, \emph{i.e.,} linear mapping network and residual network, respectively. Specifically, the fully-connected layer in neural network implements the linear mapping network. We then expand the linear mapping network to DR${}^2$-Net by adding several residual learning blocks to enhance the preliminary image. Extensive experiments demonstrate that the DR${}^2$-Net outperforms traditional iterative methods and recent deep learning-based methods by large margins at measurement rates 0.01, 0.04, 0.1, and 0.25, respectively. The code of DR${}^2$-Net has been released on: https://github.com/coldrainyht/caffe\_dr2

arXiv commentAdd the code link

References in corpus (8)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Caffe: Convolutional Architecture for Fast Feature Embedding
• Message Passing Algorithms for Compressed Sensing
• Depth Map Prediction from a Single Image using a Multi-Scale Deep Network
• Compressed Sensing of Block-Sparse Signals: Uncertainty Relations and Efficient Recovery
• Model-Based Compressive Sensing
• Compression Artifacts Removal Using Convolutional Neural Networks
• ReconNet: Non-Iterative Reconstruction of Images from Compressively Sensed Random Measurements

Cited by in corpus (31)

• Image Compressed Sensing Using Non-local Neural Network
• AutoBCS: Block-based Image Compressive Sensing with Data-driven Acquisition and Non-iterative Reconstruction
• Compressive Spectral Image Reconstruction using Deep Prior and Low-Rank Tensor Representation
• ReconResNet: Regularised Residual Learning for MR Image Reconstruction of Undersampled Cartesian and Radial Data
• Hybrid ISTA: Unfolding ISTA With Convergence Guarantees Using Free-Form Deep Neural Networks
• Deep Learning for Massive MIMO CSI Feedback
• Improving Robustness of Deep-Learning-Based Image Reconstruction
• Fast Hierarchical Deep Unfolding Network for Image Compressed Sensing
• Perceptual Compressive Sensing
• Convolutional Neural Network based Multiple-Rate Compressive Sensing for Massive MIMO CSI Feedback: Design, Simulation, and Analysis
• Efficient Physics-Based Learned Reconstruction Methods for Real-Time 3D Near-Field MIMO Radar Imaging
• Multi-resolution CSI Feedback with deep learning in Massive MIMO System
• DECONET: an Unfolding Network for Analysis-based Compressed Sensing with Generalization Error Bounds
• Learnable Descent Algorithm for Nonsmooth Nonconvex Image Reconstruction
• A 1Mb mixed-precision quantized encoder for image classification and patch-based compression
• Deep Learning Techniques for Compressive Sensing-Based Reconstruction and Inference -- A Ubiquitous Systems Perspective
• CSMCNet: Scalable Video Compressive Sensing Reconstruction with Interpretable Motion Estimation
• Scalable Deep Compressive Sensing
• MC-ISTA-Net: Adaptive Measurement and Initialization and Channel Attention Optimization inspired Neural Network for Compressive Sensing
• Improving Generalization of Sequence Encoder-Decoder Networks for Inverse Imaging of Cardiac Transmembrane Potential
• Fully Convolutional Measurement Network for Compressive Sensing Image Reconstruction
• Restricted Structural Random Matrix for Compressive Sensing
• Multi-Scale Deep Compressive Imaging
• CS-MCNet:A Video Compressive Sensing Reconstruction Network with Interpretable Motion Compensation
• Bayesian Convolutional Neural Networks for Compressed Sensing Restoration
• JR2net: A Joint Non-Linear Representation and Recovery Network for Compressive Spectral Imaging
• LSHR-Net: a hardware-friendly solution for high-resolution computational imaging using a mixed-weights neural network
• AMPA-Net: Optimization-Inspired Attention Neural Network for Deep Compressed Sensing
• Adaptive Measurement Network for CS Image Reconstruction
• QISTA-Net: DNN Architecture to Solve ${\ell }_q$-norm Minimization Problem and Image Compressed Sensing
• Multi-Scale Deep Compressive Sensing Network