摘要
LSB匹配隐写是图像隐写分析中的重点研究问题。根据图像相邻像素的相关性,提出了一种新的隐写分析算法。通过图像复原算法计算出复原图像,利用高阶Markov链模型分别对待检测图像和复原图像建模,根据LSB匹配隐写对高阶Markov链模型经验矩阵的影响,提取复原图像和待检测图像的统计特征组合成新的27维特征向量对支持向量机进行训练。实验表明提出的算法对LSB匹配隐写有较好的分析效果,特别在嵌入率低的情况下,算法具有较好的分析能力。
Detection of LSB matching steganography is an important research subject in image steganalysis. This paper pro- posed a new steganalysis algorithm based on the neighborhood correlation of pixels. It calculated restoration image through im- age restoration algorithms, and then modeled the restoration image and detection image with high order Markov chain modeled. It extracted statistical features of the restoration image and the detection image according to the effect of LSB matching on em- pirical matrix, then trained the vector support machine with the new combined 27 dimension features. Experiment results show the Proposed algorithm has a great effect on the analysis of LSB matching, especially under the condition of low embedding rate.
出处
《计算机应用研究》
CSCD
北大核心
2014年第3期846-849,共4页
Application Research of Computers