基于图像特征的隐写术载体图像的选择

Cover selection for image steganography based on image characteristics

隐写术不同于其它信息隐藏技术,信息嵌入者可以自由选择载体图像,而选择合适的载体图像有助于提高隐写术系统的不可感知性。本文通过分析载体图像的方差、复杂度、熵、直方图及其函数等图像特征和隐写术系统中的相对熵、直方度改变量等隐写术系统的性能参数之间的关系,探究图像特征是否会影响或者决定隐写术系统的性能,为隐写术载体图像选择方案提供依据。实验证明,选择合适的载体图像是提高隐写术系统不可感知性较有效的途径;在使用类LSB嵌入算法的隐写术系统中,色调、信息和纹理丰富的图像不一定是合适的载体图像,而直方图曲线变化缓慢的图像是较理想的载体图像;使用同样嵌入算法,选择合适载体图像的隐写术系统的相对熵比选择不合适载体图像低6×104多倍。

Steganography is an important branch of information hiding techniques. Different from other information hiding techniques, the embedder has the freedom to choose a cover images which results in the improvement of the undetectability of image steganography. This paper explores the relation between cover image characteristics （variance, complexity, entropy, histogram and function of histogram） and steganography performance （relative entropy and change of histogram） to find whether image characteristics affect or decide the steganography performance. The results provide basis for cover selection of image steganography. Experiments show that the selection of the proper cover image is a more effective way to improve the undetectbility of steganographic systems.