Abstract:Recent research on single-image super-resolution (SISR) using deep convolutional neural networks has made a breakthrough and achieved tremendous performance. Despite their significant progress, numerous convolutional neural networks (CNN) are limited in practical applications, owing to the requirement of the heavy computational cost of the model. This paper proposes a multi-path network for SISR, known as multi-path deep CNN with residual inception network for single image super-resolution. In detail, a residual/ResNet block with an Inception block supports the main framework of the entire network architecture. In addition, remove the batch normalization layer from the residual network (ResNet) block and max-pooling layer from the Inception block to further reduce the number of parameters to preventing the over-fitting problem during the training. Moreover, a conventional rectified linear unit (ReLU) is replaced with Leaky ReLU activation function to speed up the training process. Specifically, we propose a novel two upscale module, which adopts three paths to upscale the features by jointly using deconvolution and upsampling layers, instead of using single deconvolution layer or upsampling layer alone. The extensive experimental results on image super-resolution (SR) using five publicly available test datasets, which show that the proposed model not only attains the higher score of peak signal-to-noise ratio/structural similarity index matrix (PSNR/SSIM) but also enables faster and more efficient calculations against the existing image SR methods. For instance, we improved our method in terms of overall PSNR on the SET5 dataset with challenging upscale factor 8× as 1.88 dB over the baseline bicubic method and reduced computational cost in terms of number of parameters 62% by deeply-recursive convolutional neural network (DRCN) method.

Single Image Super-Resolution Using Deep CNN with Dense Skip Connections and Inception-ResNet

Fast and Accurate Single Image Super-Resolution Via an Energy-Aware Improved Deep Residual Network.

Lightweight Image Super-Resolution Network Using 3D Convolutional Neural Networks

Fast and Accurate Image Super Resolution by Deep CNN with Skip Connection and Network in Network

Speedy and accurate image super-resolution via deeply recursive CNN with skip connection and network in network.

Deep Neural Network for Fast and Accurate Single Image Super-Resolution via Channel-Attention-based Fusion of Orientation-aware Features

Image Super-Resolution Based on Dense Convolutional Network.

Orientation-Aware Deep Neural Network for Real Image Super-Resolution.

A Single Super-Resolution Method Via Deep Cascade Network

Symmetrical Residual Connections for Single Image Super-Resolution

Fast and Accurate Single Image Super-Resolution via Information Distillation Network

Image Superresolution Via Dense Discriminative Network.

Image Super-Resolution Via Deep Aggregation Network

Single Image Super-resolution via a Lightweight Residual Convolutional Neural Network

Image Super-Resolution Using Convolutional Neural Network with Symmetric Skip Connections

Single Image Super-Resolution Using a Polymorphic Parallel CNN

Bi-path network coupling for single image super-resolution

Deep CNN with Dense and Many-to-Many Connections for Image Super-Resolution

Image Super-Resolution Using Deep Convolutional Networks

Deeply Recursive Low- and High-Frequency Fusing Networks for Single Image Super-Resolution

Multi-Path Deep CNN with Residual Inception Network for Single Image Super-Resolution