Homomorphic encryption for dependable cloud image filtering using column permutation of pixel matrices
2026, vol.18 , no.4, pp. 103-114
Article [2026-04-10]
Cloud-based image filtering provides substantial computational benefits but exposes sensitive image data to security risks when processing is per-formed on remote systems outside the user’s control. This paper proposes a homomorphic encryption method for secure arithmetic mean image filtering based on column permutation of pixel matrices. The approach encrypts im-ages by permuting their columns before cloud transmission, delegates the computationally intensive filtering stage to remote systems, and reconstructs the filtered images locally using the inverse permutation. Experimental evaluation shows that the proposed method accelerates image filtering by one to two orders of magnitude, approaching the performance of the fastest additive masking techniques while requiring minimal user-side computation. Furthermore, it provides significantly stronger resistance to statistical reconstruction attacks, making it an efficient and secure solution for dependable cloud-based image filtering.
homomorphic encryption, cloud image filtering, column permutation, secure image processing, image streams
https://doi.org/10.59035/QWRX3585
Dimitra Georgiou, Melitine Bardi, Alireza Mirataei*, Oleksandr Markovskyi*. Homomorphic encryption for dependable cloud image filtering using column permutation of pixel matrices. International Journal on Information Technologies and Security, vol.18 , no.4, 2026, pp. 103-114. https://doi.org/10.59035/QWRX3585