Enhancing Low-resolution Image Representation Through Normalizing Flows

TL;DR

This paper introduces LR2Flow, a novel nonlinear framework combining wavelet tight frames and normalizing flows to improve low-resolution image representation, reconstruction, and processing tasks.

Contribution

It presents a new invertible neural network approach in the wavelet domain for better low-resolution image representation and reconstruction.

Findings

Effective in image rescaling, compression, and denoising

Demonstrates robustness and improved reconstruction accuracy

Validates the importance of invertible networks in wavelet domain

Abstract

Low-resolution image representation is a special form of sparse representation that retains only low-frequency information while discarding high-frequency components. This property reduces storage and transmission costs and benefits various image processing tasks. However, a key challenge is to preserve essential visual content while maintaining the ability to accurately reconstruct the original images. This work proposes LR2Flow, a nonlinear framework that learns low-resolution image representations by integrating wavelet tight frame blocks with normalizing flows. We conduct a reconstruction error analysis of the proposed network, which demonstrates the necessity of designing invertible neural networks in the wavelet tight frame domain. Experimental results on various tasks, including image rescaling, compression, and denoising, demonstrate the effectiveness of the learned representations and the robustness of the proposed framework.