摘要
研究在宿主图像内以最小比特修改量实现最大秘密信息嵌入的映射匹配隐写技术中映射关系建模问题。首先在分析既有映射匹配算法EMD算法、GLM算法基础上,进一步将二进码流的嵌入模型化为像素灰度特征码的映射匹配修正过程,并讨论了灰度特征码的映射匹配规则;随后由此提出一种采用两位灰度特征码直接映射匹配,每像素至多修正两个LSB比特位的大容量图像隐写算法:GLC3M算法。实验与理论分析结果表明,相比EMD算法、GLM算法而言,在PSNR值大于44dB的条件下,其容量比GLM算法多1倍,比EMD算法多0.7倍,并且能够有效抗击直方图差异比较与RS等隐写分析攻击。
We address the issue of mapping modeling for mapping matching steganography that can achieve the largest secret bits embedding with the least bit modifications. On the basis of analyzing the famous mapping matching algorithms such as EMD and GLM, we first further model the embedding of the binary secret bit stream as a process of gray level feature code mapping matching modification（GLC3M）, and discuss the map- ping matching rules of gray level feature codes; then we present the GLC3M algorithm of large capacity using two bits of gray level feature codes with direct mapping matching, which at most modifies two LSB bits of each pixel to implement the secret message embedding. Theoretic analysis and experimental results have shown that, in comparison with EMD and GLM, our new algorithm possesses 1 times of capacity more than that of GLM and 0.7 times more of EMD with its PSNR more than 44dB, and resists effectively typical attacks of steganalysis such as RS and visual attack with histogram comparison.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2012年第9期45-50,共6页
Journal of the China Railway Society
基金
海南大学"211工程"三期重点学科建设项目专项资金
海南大学博士科研启动经费专项资金(kyqd1109)
关键词
灰度特征码
映射匹配修正
隐写术
隐写分析攻击
gray level feature code
mapping matching modification
steganography
steganalysis
