基于JPEG图象隐写算法的统计不可检测性比较分析

A Comparing Analysis of Statistical Undetectability for JPEG based Steganograghy

本研究选择现今几种基于JPEG图象的隐写算法,并通过测试实验,比较分析各算法在统计不可检测性的表现。同时研究了隐写算法设计因素的原则,包括矩阵潜入、适应性内容依赖选择信道、边信息等对嵌入的影响。实验结果表明,要使JPEG会阶图象的统计不可检测性表现最好,嵌入比率应该在0.05比特每非0DCT直流系数。

This research selects current best steganograghy methods, then compares and analyzes the performance on statistical undetectability. Additionally, by testing selected steganographic algorithms this research evaluates the influence of specific design elements and principles, such as matrix embedding, adaptive content-dependent selection channels, and side information at the sender. From the experiments, it can be concluded that the average steganographic capacity of JPEG images is approximately 0.05 bits per nonzero AC DCT coefficient.