Modification in spatial, extraction from transform: A new approach for JPEG steganography

Steganography is a choice in order to have a secret communication and it can be achieved by embedding a message inside a carrier object. JPEG as a common image format is a good target for steganography. In JPEG steganography, due to compression artifacts, it is not possible to embed data in pixels. As a result, most of the prevalent steganography methods for JPEG tend to embed data in JPEG coefficients and thus, the modification - on the side of sender - and extraction of data - on the side of recipient - will be made in the same domain. This paper aims at introducing a new method for JPEG steganography in which, modification and extraction of data is performed in different domains. In the proposed method, some intentional changes in spatial domain is made on the side of sender. These changes develop a specific pattern in JPEG coefficients and the recipient should extract data from JPEG coefficients. Most of JPEG steganography methods use a key in order to decentralize modification artifacts and therefore avoid detection. In the proposed method, the need for steganography secret key is removed. The proposed method also opposes the common idea that fewer changes in JPEG coefficients lead to less detectability. The experimental results including comparison of the proposed method with one of the outstanding methods for JPEG steganography show that the proposed method alters more JPEG coefficients and it is yet less detectable.