DCT-in-DCT: A Novel Steganography Scheme for Enhanced Payload Extraction Quality

A common practice of many steganography schemes is to replace the high frequency insignificant Discrete Cosine Transform (DCT) coefficients with secret data. The replaced secret data, however, needs to be scaled appropriately before replacing the insignificant DCT coefficients. The secret is typically scaled down to closely match the replaced insignificant coefficients of the cover. State of the art techniques show that using multiple scales to adaptively scale the secret improves the stego quality without compromising the embedding capacity. Scaling can also potentially lead to degradation of the extracted image quality due to high scaling factors. Therefore, this paper proposes a novel DCT-in-DCT-Adaptive Scaling technique that achieves higher payload imperceptibility without compromise in the stego quality for a fixed embedding capacity. The proposed technique utilizes DCT transform to find the DCT magnitude of the secret before adaptively scaling the secret and replacing the scaled secret with high frequency insignificant DCT coefficients of the cover image. Results compared with Spatial-in-DCT-Adaptive Scaling show improvement of up to 20.25 dB in the extracted quality of the payload.