Toward Real World Stereo Image Super-Resolution via Hybrid Degradation Model and Discriminator for Implied Stereo Image Information (2023-12-13T00:00:00.000000Z)

TL;DR

A novel approach is proposed that integrates a implicit stereo information discriminator and a hybrid degradation model that ensures effective enhancement while preserving disparity consistency, and demonstrates impressive performance on synthetic and real datasets.

Abstract

Real-world stereo image super-resolution has a significant influence on enhancing the performance of computer vision systems. Although existing methods for single-image super-resolution can be applied to improve stereo images, these methods often introduce notable modifications to the inherent disparity, resulting in a loss in the consistency of disparity between the original and the enhanced stereo images. To overcome this limitation, this paper proposes a novel approach that integrates a implicit stereo information discriminator and a hybrid degradation model. This combination ensures effective enhancement while preserving disparity consistency. The proposed method bridges the gap between the complex degradations in real-world stereo domain and the simpler degradations in real-world single-image super-resolution domain. Our results demonstrate impressive performance on synthetic and real datasets, enhancing visual perception while maintaining disparity consistency. The complete code is available at the following \href{https://github.com/fzuzyb/SCGLANet}{link}.

Authors

Qinquan Gao

2 papers

Yuanbo Zhou

1 papers

Yuyang Xue

1 papers

TL;DR

Abstract

Authors

References68 items

NTIRE 2023 Challenge on Stereo Image Super-Resolution: Methods and Results

Stereo Cross Global Learnable Attention Module for Stereo Image Super-Resolution

Global Learnable Attention for Single Image Super-Resolution

Semantic Guided Long Range Stereo Depth Estimation for Safer Autonomous Vehicle Applications

Heat Transfer-Inspired Network for Image Super-Resolution Reconstruction

Metric Learning based Interactive Modulation for Real-World Super-Resolution

NAFSSR: Stereo Image Super-Resolution Using NAFNet

Efficient and Degradation-Adaptive Network for Real-World Image Super-Resolution

Learning the Degradation Distribution for Blind Image Super-Resolution

Self-Supervised Online Learning for Safety-Critical Control using Stereo Vision

Generative Adversarial Networks

Perception-Oriented Stereo Image Super-Resolution

RAFT-Stereo: Multilevel Recurrent Field Transforms for Stereo Matching

Transformer for Single Image Super-Resolution

SwinIR: Image Restoration Using Swin Transformer

Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data

Cross View Capture for Stereo Image Super-Resolution

Feedback Network for Mutually Boosted Stereo Image Super-Resolution and Disparity Estimation

Designing a Practical Degradation Model for Deep Blind Image Super-Resolution

Cross Parallax Attention Network for Stereo Image Super-Resolution

Optimizing depth perception in virtual and augmented reality through gaze-contingent stereo rendering

Interpreting Super-Resolution Networks with Local Attribution Maps

Deep Stereoscopic Image Super-Resolution via Interaction Module

Symmetric Parallax Attention for Stereo Image Super-Resolution

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

Real-World Super-Resolution via Kernel Estimation and Noise Injection

Disparity-Aware Domain Adaptation in Stereo Image Restoration

Learning Texture Transformer Network for Image Super-Resolution

Iterative Network for Image Super-Resolution

Stereoscopic Image Super-Resolution with Stereo Consistent Feature

A Stereo Attention Module for Stereo Image Super-Resolution

Fine-Grained Attention and Feature-Sharing Generative Adversarial Networks for Single Image Super-Resolution

Unsupervised Learning for Real-World Super-Resolution

RankSRGAN: Generative Adversarial Networks With Ranker for Image Super-Resolution

Searching for MobileNetV3

Blind Super-Resolution With Iterative Kernel Correction

Flickr1024: A Large-Scale Dataset for Stereo Image Super-Resolution

Learning Parallax Attention for Stereo Image Super-Resolution

2018 PIRM Challenge on Perceptual Image Super-resolution

ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks

Deep Learning for Single Image Super-Resolution: A Brief Review

Image Super-Resolution Using Very Deep Residual Channel Attention Networks

Enhancing the Spatial Resolution of Stereo Images Using a Parallax Prior

Residual Dense Network for Image Super-Resolution

Spectral Normalization for Generative Adversarial Networks

The Unreasonable Effectiveness of Deep Features as a Perceptual Metric

Image Super-Resolution Using Dense Skip Connections

Enhanced Deep Residual Networks for Single Image Super-Resolution

Attention is All you Need

Adversarial PoseNet: A Structure-Aware Convolutional Network for Human Pose Estimation

DSLR-Quality Photos on Mobile Devices with Deep Convolutional Networks

Learning a No-Reference Quality Metric for Single-Image Super-Resolution

Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

SGDR: Stochastic Gradient Descent with Warm Restarts

Accelerating the Super-Resolution Convolutional Neural Network

Layer Normalization

Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network

Accurate Image Super-Resolution Using Very Deep Convolutional Networks

Adam: A Method for Stochastic Optimization

Very Deep Convolutional Networks for Large-Scale Image Recognition

High-Resolution Stereo Datasets with Subpixel-Accurate Ground Truth

Are we ready for autonomous driving? The KITTI vision benchmark suite

SRDRL: A Blind Super-Resolution Framework With Degradation Reconstruction Loss

Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

is with the Beijing Radio and TV Station, Beijing, 100022, China

is with the College of Physics and Information Engineering, Fuzhou University, Fuzhou, 350108, China

Graduate Student Member

is with the Imperial Vision Technology, Fuzhou, 350108, China

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL