Embedding information into JPEG images with distortion masking in frequency domain

Abstract

Steganography allows to ensure the confidentiality of information by organizing covert data transmission channels. However, the effectiveness of steganographic information protection directly depends on the invisibility of a secret message, both for the human eye and for steganalysis methods. The paper proposes an approach that allows solving the problem of vulnerability of the popular QIM embedding method to statistical steganalysis. For this, it is proposed to use a variable quantization step, which is adaptively selected for each block of the JPEG cover image. The experimental results demonstrate an increase in the security level of steganographic embedding due to the application of the proposed approach.