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.

提出了一种适用于彩色图像隐写术的自适应模糊推理方法，该方法考虑了图像复杂度的影响，如像素相似度，像素亮度和色彩灵敏度。将模糊推理系统设计为分类器，该系统将封面图像的特征用作其清晰的输入值，并生成与图像子类的有效负载相对应的语义概念。此外，根据模糊推理系统的输出以及人眼对R，G，B颜色分量的敏感性，使用最低有效位替换来自适应地隐藏数据。将混沌方法和随机序列加扰应用于秘密消息以生成防止来自攻击者的秘密消息的随机序列。所提出的方法从人的视觉系统中隐藏了大量具有隐秘质量的数据，并保证了通信的机密性。实验结果表明，较好的均方误差，峰值信噪比，结构相似性和有效负载，证明了该方法比某些最新技术能产生更好的性能。该方法的鲁棒性通过RS隐写分析和像素差异直方图分析进行了测试。