Synthesis and Evaluation of Improved Reference Matrix Models for High Capacity Image Steganography

Abstract

This paper proposes two new reference matrix based steganographic methods called Grid Window Method (GWM) and Exploited Grid Window Method (EGWM). The goal of image steganography is to embed secret information within ingenuous cover images to protect the former from third-party adversarial attacks. The primary aim of this paper is to increase the size of the location map corresponding to the reference matrix which consequently increases the number of secrets digits that can be hidden within a single cover image pixel. Reference matrices are often constructed using turtle shells (hexagons) whereas the proposed grid windows are square-shaped. The number of secret digits hidden per cover-pixel pair is four in GWM and five in EGWM. GWM offers similar embedding capacity when compared to existing turtle shell methods (2 bpp) and EGWM surpasses these results by 25% (2.5 bpp). The parameters PSNR and SSIM for these models are also compared and evaluated.