Abstract:Deep Learning (DL) based Compressed Sensing (CS) has been applied for better performance of image reconstruction than traditional CS methods. However, most existing DL methods utilize the block-by-block measurement and each measurement block is restored separately, which introduces harmful blocking effects for reconstruction. Furthermore, the neuronal receptive fields of those methods are designed to be the same size in each layer, which can only collect single-scale spatial information and has a negative impact on the reconstruction process. This paper proposes a novel framework named Multi-scale Dilated Convolution Neural Network (MsDCNN) for CS measurement and reconstruction. During the measurement period, we directly obtain all measurements from a trained measurement network, which employs fully convolutional structures and is jointly trained with the reconstruction network from the input image. It needn't be cut into blocks, which effectively avoids the block effect. During the reconstruction period, we propose the Multi-scale Feature Extraction (MFE) architecture to imitate the human visual system to capture multi-scale features from the same feature map, which enhances the image feature extraction ability of the framework and improves the performance of image reconstruction. In the MFE, there are multiple parallel convolution channels to obtain multi-scale feature information. Then the multi-scale features information is fused and the original image is reconstructed with high quality. Our experimental results show that the proposed method performs favorably against the state-of-the-art methods in terms of PSNR and SSIM.

Efficient Multiple Frame Images Recovery Based on Distributed Compressed Sensing

Image Reconstruction Via Small Block-Based Compressed Sensing and Discrete Legendre Polynomials

Compressed Sensing Fourier Single Pixel Imaging Algorithm Based on Joint Discrete Gradient and Non-Local Self-Similarity Priors

Image reconstruction based on back propagation learning in Compressed Sensing theory

Simultaneous Sparse Approximation and Common Component Extraction using Fast Distributed Compressive Sensing

Image compressive-sensing recovery using structured laplacian sparsity in DCT domain and multi-hypothesis prediction

Video Compressive Sensing Reconstruction Via Reweighted Residual Sparsity

Multi-slice and Multi-Frame Image Reconstruction by Predictive Compressed Sensing.

Learned Distributed Image Compression with Multi-Scale Patch Matching in Feature Domain

Residual image recovery method based on the dual-camera design of a compressive hyperspectral imaging system

Adaptive High-Resolution Imaging Method Based on Compressive Sensing

Image Multiple Description Coding Method Based on Interleaving Extraction and Block Compressive Sensing Strategy

Structural Group Sparse Representation for Image Compressive Sensing Recovery

Image Compressed Sensing with Multi-scale Dilated Convolutional Neural Network

Image Compressive Sensing Recovery Via Collaborative Sparsity

Compressed Multi-view Imaging with Joint Reconstruction

Image Compressed Sensing Reconstruction with 3D Transform Domain Collaborative Filtering

Statistical Prior Aided Separate Compressed Image Sensing for Green Internet of Multimedia Things

High-Quality Image Compressed Sensing and Reconstruction with Multi-scale Dilated Convolutional Neural Network

Compressive-Sensed Image Coding Via Stripe-based DPCM

Bioactive compounds from marine mussels and their effects on human health.