Saved in:
Bibliographic Details
Main Authors: Feng, Xue, Strohmer, Thomas
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2404.08900
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929313264173056
author Feng, Xue
Strohmer, Thomas
author_facet Feng, Xue
Strohmer, Thomas
contents Autoencoders are important generative models that, among others, have the ability to interpolate image sequences. However, interpolated images are usually not semantically meaningful.In this paper, motivated by dynamic optimal transport, we consider image interpolation as a mass transfer problem and propose a novel regularization term to penalize non-smooth and unrealistic changes in the interpolation result. Specifically, we define the path energy function for each path connecting the source and target images. The autoencoder is trained to generate the $L^2$ optimal transport geodesic path when decoding a linear interpolation of their latent codes. With a simple extension, this model can handle complicated environments, such as allowing mass transfer between obstacles and unbalanced optimal transport. A key feature of the proposed method is that it is physics-driven and can generate robust and realistic interpretation results even when only very limited training data are available.
format Preprint
id arxiv_https___arxiv_org_abs_2404_08900
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Improving Autoencoder Image Interpolation via Dynamic Optimal Transport
Feng, Xue
Strohmer, Thomas
Optimization and Control
Autoencoders are important generative models that, among others, have the ability to interpolate image sequences. However, interpolated images are usually not semantically meaningful.In this paper, motivated by dynamic optimal transport, we consider image interpolation as a mass transfer problem and propose a novel regularization term to penalize non-smooth and unrealistic changes in the interpolation result. Specifically, we define the path energy function for each path connecting the source and target images. The autoencoder is trained to generate the $L^2$ optimal transport geodesic path when decoding a linear interpolation of their latent codes. With a simple extension, this model can handle complicated environments, such as allowing mass transfer between obstacles and unbalanced optimal transport. A key feature of the proposed method is that it is physics-driven and can generate robust and realistic interpretation results even when only very limited training data are available.
title Improving Autoencoder Image Interpolation via Dynamic Optimal Transport
topic Optimization and Control
url https://arxiv.org/abs/2404.08900