MRAS: A matching robust adaptive steganography scheme for JPEG images over social networking platforms

Abstract

Steganography on social networking platforms (SNPs) faces significant challenges due to JPEG recompression, which distorts hidden information. Traditional techniques lack robustness against these recompression-induced losses. To address this, we propose a novel Matching Robust Adaptive Steganography (MRAS) scheme for JPEG images shared over SNPs. MRAS enhances robustness through: (i) a preprocessing stage to stabilize DCT coefficients, (ii) a lattice-based adaptive embedding strategy for resilience, and (iii) a postprocessing stage to mitigate embedding-induced distortions. Experimental results on benchmark images demonstrate that MRAS offers a better tradeoff among robustness, undetectability, imperceptibility, and payload capacity compared to state-of-the-art methods. Furthermore, real-world testing on LinkedIn confirms its practicality and effectiveness for secure communication in realistic settings.