结合融合特征与特征映射的空域图像隐写分析

Steganalysis of spatial image combining fusion features and feature mapping

为了更好地捕获隐写术对图像统计特征的改变,提高对隐写图像的检测率并解决特征映射问题,提出结合融合特征与特征映射的隐写分析方法,提取融合特征,更全面地捕获隐写算法对载体图像统计特征的扰动。同时,提出结合PCA的特征映射,以解决图像数小于特征维数时不能直接投影的问题。然后,对融合特征进行结合PCA的近似映射,用于隐写分析。实验证明:该方法有效地提升了对隐写图像的检测率。

In order to better capture the changes of steganography to the statistical characteristics of images,improve the detection rate of steganographic images and solve the problem of feature mapping,a steganalysis method combining fusion features and feature mapping is proposed to extract fusion features and capture more comprehensively the perturbation of the steganographic algorithm to the statistical characteristics of the carrier image.And a feature map combined with PCA is proposed to solve the problem of direct projection when the number of images is less than the feature dimension.The fused features are then subjected to approximate mapping combined with PCA for steganalysis.Experiments show that this method can effectively improve the detection rate of steganographic images.