A Parallel Region-Adaptive Differential Privacy Framework for Image Pixelization
Ming Liu · 2025
Read A Parallel Region-Adaptive Differential Privacy Framework for Image Pixelization by Ming Liu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-04261.txt.
Description
Ming Liu, writing alone, presents a parallel, region-adaptive framework for private image pixelization, where grid noise gives formal privacy guarantees for visual data. High-resolution sensing plus foundation models raise re-identification risk, while classic pixelization struggles with fidelity, scale, and real-time speed at once.
Grid sizes and noise scales adapt to regional detail, and graphics chip parallelism speeds the process well beyond the classic baseline. Only key noisy stats are stored, cutting space sharply. Formal analysis uses the Laplace mechanism and parallel composition. Tests on three video datasets showed favourable privacy-utility trade-offs, and a face re-identification attack test on a celebrity set confirmed protection in practice.
Catalog record
| CREATOR | Ming Liu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 73 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-04261