Effective Invertible Arbitrary Image Rescaling

TL;DR

This paper introduces a novel invertible neural network architecture that enables arbitrary image rescaling, including asymmetric scales, with a single model, achieving state-of-the-art results without increasing complexity.

Contribution

The work proposes a simple, effective invertible rescaling network that handles arbitrary and asymmetric scales with one model, using innovative encoding and channel splitting techniques.

Findings

Achieves state-of-the-art bidirectional arbitrary rescaling performance.

Handles asymmetric scale factors effectively with the same network architecture.

Maintains high perceptual quality in low-resolution outputs.

Abstract

Great successes have been achieved using deep learning techniques for image super-resolution (SR) with fixed scales. To increase its real world applicability, numerous models have also been proposed to restore SR images with arbitrary scale factors, including asymmetric ones where images are resized to different scales along horizontal and vertical directions. Though most models are only optimized for the unidirectional upscaling task while assuming a predefined downscaling kernel for low-resolution (LR) inputs, recent models based on Invertible Neural Networks (INN) are able to increase upscaling accuracy significantly by optimizing the downscaling and upscaling cycle jointly. However, limited by the INN architecture, it is constrained to fixed integer scale factors and requires one model for each scale. Without increasing model complexity, a simple and effective invertible arbitrary rescaling network (IARN) is proposed to achieve arbitrary image rescaling by training only one model in this work. Using innovative components like position-aware scale encoding and preemptive channel splitting, the network is optimized to convert the non-invertible rescaling cycle to an effectively invertible process. It is shown to achieve a state-of-the-art (SOTA) performance in bidirectional arbitrary rescaling without compromising perceptual quality in LR outputs. It is also demonstrated to perform well on tests with asymmetric scales using the same network architecture.