Secure high payload steganography: A model-based approach

Abstract

The strong energy compaction property of the Discrete Cosine Transform (DCT) has inspired DCT-based steganography techniques to embed secret data in large areas of the insignificant DCT coefficients. Since these coefficients are noncritical, conventional DCT-based steganography techniques replace these insignificant coefficients with the secret data. This basic idea has enabled DCT-based steganography techniques to hide a substantial amount of secret data while maintaining acceptable stego fidelity. However, because these insignificant DCT coefficients do still contribute to the overall quality of the resulting stego image in the spatial domain, there will be an advantage in retaining them as much as possible. Thus, to optimize the stego quality, this paper proposes a novel hiding strategy where the general structure of the insignificant DCT coefficients is retained. A model-based technique is developed using polynomial surface-fitting of the insignificant DCT coefficients, and the secret data is superimposed on the polynomial-modeled coefficients, thus producing a stego image that does not compromise the overall structure of the original cover image. Experimental results demonstrating improvements in the stego quality and security levels of the proposed steganography technique in comparison to state-of-the-art schemes are presented.