Rotation Equivariant Arbitrary-scale Image Super-Resolution

2025-08-07

AI Summary

To address structural artifacts arising from geometric pattern distortion in Arbitrary-Scale Image Super-Resolution (ASISR), this paper proposes the first end-to-end rotation-equivariant ASISR framework. Methodologically, we design a deep encoder and an Implicit Neural Representation (INR) module both inherently rotation-equivariant, integrating continuous implicit function modeling with equivariant convolutions to achieve equivariant mapping across arbitrary scale factors. A rigorous theoretical analysis establishes a tight bound on the model’s equivariance error. Experiments demonstrate that our approach significantly suppresses geometric artifacts on both synthetic and real-world datasets, substantially improving structural fidelity—particularly for textures, edges, and shapes. Moreover, its modular architecture enables plug-and-play integration to enhance the rotation-equivariance robustness of existing super-resolution models without architectural modification.

Technology Category

• Computer Vision: Representation Learning for Vision
• Machine Learning: Representation Learning
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

The arbitrary-scale image super-resolution (ASISR), a recent popular topic in computer vision, aims to achieve arbitrary-scale high-resolution recoveries from a low-resolution input image. This task is realized by representing the image as a continuous implicit function through two fundamental modules, a deep-network-based encoder and an implicit neural representation (INR) module. Despite achieving notable progress, a crucial challenge of such a highly ill-posed setting is that many common geometric patterns, such as repetitive textures, edges, or shapes, are seriously warped and deformed in the low-resolution images, naturally leading to unexpected artifacts appearing in their high-resolution recoveries. Embedding rotation equivariance into the ASISR network is thus necessary, as it has been widely demonstrated that this enhancement enables the recovery to faithfully maintain the original orientations and structural integrity of geometric patterns underlying the input image. Motivated by this, we make efforts to construct a rotation equivariant ASISR method in this study. Specifically, we elaborately redesign the basic architectures of INR and encoder modules, incorporating intrinsic rotation equivariance capabilities beyond those of conventional ASISR networks. Through such amelioration, the ASISR network can, for the first time, be implemented with end-to-end rotational equivariance maintained from input to output. We also provide a solid theoretical analysis to evaluate its intrinsic equivariance error, demonstrating its inherent nature of embedding such an equivariance structure. The superiority of the proposed method is substantiated by experiments conducted on both simulated and real datasets. We also validate that the proposed framework can be readily integrated into current ASISR methods in a plug & play manner to further enhance their performance.

Problem

• Achieve arbitrary-scale high-resolution recovery from low-resolution images
• Address warped geometric patterns in low-resolution image upscaling
• Embed rotation equivariance to maintain structural integrity in super-resolution

Innovation

• Rotation equivariant ASISR network design
• Redesigned INR and encoder modules
• End-to-end rotational equivariance maintenance