Abstract: For the existing steganography, it is hard to extract secret information without original cover-image and easy to be detected by steganalysts when the hiding capacity is high. To solve this problem, a new scheme of steganography based on fractal graph was proposed. In this scheme, firstly, Black-and-White Fractal Graph (BWFG) was created by utilizing affine transformation and fractal iterated function system. Then the BWFG was transformed to Black-and-White Pixel Image (BWPI) based on the idea of coordinate transformation. At last, the BWPI was divided into several non-overlapping blocks and the positions of black and white pixels in each block were altered to hide the secret information, generating stego-image. The receiver could create the cover-image by utilizing the parameters of affine transformation and times of iteration, and extract secret information by comparing the difference of pixels in corresponding blocks. Theoretical analysis and simulation experiments show that, compared with the information hiding algorithm in frequency domain, the proposed scheme has good imperceptibility and high hiding capacity, and can resist steganalysis based on image features and transform domain coefficient change.

Key words: fractal graph, information hiding, imperceptibility, hiding capacity, steganalysis

摘要： 针对现有信息隐藏提取得不到原始载体图像使提取困难、隐藏容量高又易被隐密分析攻破的问题，提出以分形图形为载体的信息隐藏方法。首先，利用数学仿射变换及分形迭代函数系统生成黑白的分形图形（BWFG）；然后，根据坐标变换的思想，将分形图形转化为黑白像素图像（BWPI）；最后，将像素图像分成不重叠的小块，通过调整小块内黑白像素的位置来隐藏秘密信息，形成含密图像。知道仿射变换参数及迭代次数的秘密信息接收方，首先生成原始载体分形图像，然后通过对比两幅图像相应小块黑白像素差异，即可提取隐藏的秘密信息。理论分析和仿真实验表明，与频率域信息隐藏算法相比，提出的算法隐蔽性强、隐藏容量高，能抵抗基于图像特征和变换域系数变化的隐密分析。

关键词: 分形图形, 信息隐藏, 隐蔽性, 隐藏容量, 隐密分析