Leveraging Cold Diffusion for the Decomposition of Identically Distributed Superimposed Images

With the growing adoption of Deep Learning for imaging tasks in biometrics and healthcare, it becomes increasingly important to ensure privacy when using and sharing images of people. Several works enable privacy-preserving image sharing by anonymizing the images so that the corresponding individual...

Full description

Saved in:
Bibliographic Details
Main Authors: Helena Montenegro, Jaime S. Cardoso
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Open Journal of Signal Processing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11054277/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849319203249586176
author Helena Montenegro
Jaime S. Cardoso
author_facet Helena Montenegro
Jaime S. Cardoso
author_sort Helena Montenegro
collection DOAJ
description With the growing adoption of Deep Learning for imaging tasks in biometrics and healthcare, it becomes increasingly important to ensure privacy when using and sharing images of people. Several works enable privacy-preserving image sharing by anonymizing the images so that the corresponding individuals are no longer recognizable. Most works average images or their embeddings as an anonymization technique, relying on the assumption that the average operation is irreversible. Recently, cold diffusion models, based on the popular denoising diffusion probabilistic models, have succeeded in reversing deterministic transformations on images. In this work, we leverage cold diffusion to decompose superimposed images, empirically demonstrating that it is possible to obtain two or more identically-distributed images given their average. We propose novel sampling strategies for this task and show their efficacy on three datasets. Our findings highlight the risks of averaging images as an anonymization technique and argue for the use of alternative anonymization strategies.
format Article
id doaj-art-e3c91537f8d1484e962480cb12affc7f
institution Kabale University
issn 2644-1322
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Open Journal of Signal Processing
spelling doaj-art-e3c91537f8d1484e962480cb12affc7f2025-08-20T03:50:32ZengIEEEIEEE Open Journal of Signal Processing2644-13222025-01-01678479410.1109/OJSP.2025.358396311054277Leveraging Cold Diffusion for the Decomposition of Identically Distributed Superimposed ImagesHelena Montenegro0https://orcid.org/0000-0001-6237-3011Jaime S. Cardoso1https://orcid.org/0000-0002-3760-2473Faculty of Engineering, University of Porto, Porto, PortugalFaculty of Engineering, University of Porto, Porto, PortugalWith the growing adoption of Deep Learning for imaging tasks in biometrics and healthcare, it becomes increasingly important to ensure privacy when using and sharing images of people. Several works enable privacy-preserving image sharing by anonymizing the images so that the corresponding individuals are no longer recognizable. Most works average images or their embeddings as an anonymization technique, relying on the assumption that the average operation is irreversible. Recently, cold diffusion models, based on the popular denoising diffusion probabilistic models, have succeeded in reversing deterministic transformations on images. In this work, we leverage cold diffusion to decompose superimposed images, empirically demonstrating that it is possible to obtain two or more identically-distributed images given their average. We propose novel sampling strategies for this task and show their efficacy on three datasets. Our findings highlight the risks of averaging images as an anonymization technique and argue for the use of alternative anonymization strategies.https://ieeexplore.ieee.org/document/11054277/Image anonymizationsuperimposed image decompositiondiffusion models
spellingShingle Helena Montenegro
Jaime S. Cardoso
Leveraging Cold Diffusion for the Decomposition of Identically Distributed Superimposed Images
IEEE Open Journal of Signal Processing
Image anonymization
superimposed image decomposition
diffusion models
title Leveraging Cold Diffusion for the Decomposition of Identically Distributed Superimposed Images
title_full Leveraging Cold Diffusion for the Decomposition of Identically Distributed Superimposed Images
title_fullStr Leveraging Cold Diffusion for the Decomposition of Identically Distributed Superimposed Images
title_full_unstemmed Leveraging Cold Diffusion for the Decomposition of Identically Distributed Superimposed Images
title_short Leveraging Cold Diffusion for the Decomposition of Identically Distributed Superimposed Images
title_sort leveraging cold diffusion for the decomposition of identically distributed superimposed images
topic Image anonymization
superimposed image decomposition
diffusion models
url https://ieeexplore.ieee.org/document/11054277/
work_keys_str_mv AT helenamontenegro leveragingcolddiffusionforthedecompositionofidenticallydistributedsuperimposedimages
AT jaimescardoso leveragingcolddiffusionforthedecompositionofidenticallydistributedsuperimposedimages