Multiresolution Neural Networks for Imaging

TL;DR

This paper introduces MR-Net, a multiresolution neural network architecture for imaging that offers continuous, compact, and efficient representations, demonstrated through applications like texture magnification, minification, and antialiasing.

Contribution

The paper proposes a novel multiresolution neural network architecture, MR-Net, that is continuous in space and scale, and demonstrates its effectiveness in various imaging tasks.

Findings

Effective multiresolution image representation

Applications to texture magnification and minification

Improved antialiasing results

Abstract

We present MR-Net, a general architecture for multiresolution neural networks, and a framework for imaging applications based on this architecture. Our coordinate-based networks are continuous both in space and in scale as they are composed of multiple stages that progressively add finer details. Besides that, they are a compact and efficient representation. We show examples of multiresolution image representation and applications to texturemagnification, minification, and antialiasing. This document is the extended version of the paper [PNS+22]. It includes additional material that would not fit the page limitations of the conference track for publication.