An adaptive fuzzy inference approach for color image steganography

This paper proposes an adaptive fuzzy inference approach for color image steganography, taking into account the influence of image complexity such as pixel similarity, pixel brightness and color sensitivity. A fuzzy inference system is designed as a classifier which adopts the features of the cover image as its crisp input values and produces semantic concepts corresponding to the payload of image sub-classes. Furthermore, least significant bit substitution is used to hide the data adaptively according to the output of fuzzy inference system and the human eye sensitivity to the R, G, B color components. A chaotic method and random sequence scrambling are applied to the secret message to generate the random sequence which prevents the secret message from attackers. The proposed method hides a large amount of data with good quality of stego-image from the human visual system and guarantees the confidentiality in the communication. Experimental results show better mean square error, peak signal-to-noise ratio, structural similarity and payload, verifying that the proposed method can yield better performance than some state-of-the-art works. The robustness of the method is tested by RS steganalysis and pixel difference histogram analysis.