Abstract:Real-SR endeavors to produce high-resolution images with rich details while mitigating the impact of multiple degradation factors. Although existing methods have achieved impressive achievements in detail recovery, they still fall short when addressing regions with complex gradient arrangements due to the intensity-based linear weighting feature extraction manner. Moreover, the stochastic artifacts introduced by degradation cues during the imaging process in real LR increase the disorder of the overall image details, further complicating the perception of intrinsic gradient arrangement. To address these challenges, we innovatively introduce kernel-wise differential operations within the convolutional kernel and develop several learnable directional gradient convolutions. These convolutions are integrated in parallel with a novel linear weighting mechanism to form an Adaptive Directional Gradient Convolution (DGConv), which adaptively weights and fuses the basic directional gradients to improve the gradient arrangement perception capability for both regular and irregular textures. Coupled with DGConv, we further devise a novel equivalent parameter fusion method for DGConv that maintains its rich representational capabilities while keeping computational costs consistent with a single Vanilla Convolution (VConv), enabling DGConv to improve the performance of existing super-resolution networks without incurring additional computational expenses. To better leverage the superiority of DGConv, we further develop an Adaptive Information Interaction Block (AIIBlock) to adeptly balance the enhancement of texture and contrast while meticulously investigating the interdependencies, culminating in the creation of a DGPNet for Real-SR through simple stacking. Comparative results with 15 SOTA methods across three public datasets underscore the effectiveness and efficiency of our proposed approach.

Neural Super-Resolution in Real-Time Rendering Using Auxiliary Feature Enhancement

Neural supersampling for real-time rendering

Neural Super-Resolution for Real-time Rendering with Radiance Demodulation

Deep Fourier-based Arbitrary-scale Super-resolution for Real-time Rendering

Neural Foveated Super-Resolution for Real-time VR Rendering

Neural Global Illumination Via Superposed Deformable Feature Fields

Efficient Deep Models for Real-Time 4K Image Super-Resolution. NTIRE 2023 Benchmark and Report.

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

Efficient neural supersampling on a novel gaming dataset

GPU Accelerated High-Quality Video/image Super-Resolution

Exploring Real-Time Super-Resolution Generative Adversarial Networks

Auxiliary Features-Guided Super Resolution for Monte Carlo Rendering

Super-Resolution Neural Radiance Field via Learning High Frequency Details for High-Fidelity Novel View Synthesis

An Improved Image Super Resolution and Its Parallel Implementation Based on CUDA.

Efficient Real-world Image Super-Resolution Via Adaptive Directional Gradient Convolution

A Real-Time Super-Resolution Method Based on Convolutional Neural Networks.

Real-world video superresolution enhancement method based on the adaptive down-sampling model

Neural Volume Super-Resolution

Gradient-Guided and Multi-Scale Feature Network for Image Super-Resolution

A Novel Image Super-Resolution Approach for Industrial Product Visual Enhancement

Efficient Image Super-Resolution Integration