Ma Lihong, Lv Xianming, Gao Xiaoman. Most significant bits probability arithmetic decoding for improved image steganography[J]. Journal of Image and Graphics, 2012, 17(5): 621-627. DOI: 10.11834/jig.20120503.
Most significant bits probability arithmetic decoding for improved image steganography
提出一种基于MSBs(高位有效位)概率算术解码的DCT(离散余弦变换)域YASS (yet another steganographic scheme)密写改进算法。为了克服YASS算法因量化索引调制(QIM)嵌入特性造成的易攻击缺陷
在YASS消息嵌入中用原型信号MSBs概率分布约束的算术解码
使消息嵌入流的解码序列具有与原型信号一致的边缘分布
在保持YASS算法高随机性嵌入优点的同时
能使密写前后一阶分布直方图之间的差异最小化
有效保证了密写的安全性。实验结果表明
与YASS对比
该方法使系数分布直方图的平均变化量更小
其平均失真为6.3%
相比YASS的平均失真6.8%
减少了0.5%
改进算法的抗隐写分析能力比后者更强
是一种有效的密写算法。
Abstract
To overcome the attacking-prone drawbacks of quantization index modulation(QIM) in YASS(yet another steganographic scheme)
an improved YASS algorithm based on MSBs(most significant bits)arithmetic decoding in DCT domain is proposed.First
a secret sequence is viewed as an arithmetic coding stream and recovered by arithmetic decoding with the MSBs probability of a cover signal distributed
thus the codec reset results in a same marginal distribution between decoding stream and the cover.Then
the decoded bits are reembedded as the convention.Combined with the highly random embedding of YASS
the proposed scheme has a higher security with the first order distribution historgram between the cover and the stego minimized.Experiments demonstrate that the proposed method outperforms YASS both in capability of anti-steg-analysis and the performance of coefficient distribution historgram
which is 6.8% for YASS and 6.3% for the improved one in average. It is an effective steganographic algorithm.