摘要
传统修改式信息隐藏无法抵抗密写分析的检测,纹理构造式信息隐藏只能生成简单质地的纹理,不能对秘密信息进行有效掩盖,纹理拼接式信息隐藏,秘密信息与样本小块之间存在着固定的映射关系以及码块间的区别特征,会导致信息泄露,降低其安全性。针对以上问题,提出了一种结合随机映射和改进缝合线的纹理合成隐藏方法。首先将样本图像划分为样本小块并依据样本小块均值划分为不同类别,然后通过建立起秘密信息分段和样本小块类别间的随机映射关系来编码秘密信息并将其放置在空白图像的特定位置上,最后按改进缝合线纹理合成方法将放置小块与相邻小块进行纹理合成并生成含密纹理。实验结果表明,该方法可通过合成纹理来对秘密信息进行掩盖,秘密信息分段和编码样本小块不一一对应,该方法完全依赖于用户密钥且具备一定的抗攻击能力。
Traditional modified information hiding cannot resist detections from steganographic analysis.Texture constructive information hiding can only generate simple texture and it cannot cover secret information effectively.In texture splicing information hiding,it leads to information leakage and poor security due to the fixed mapping relationship between secret information segments and sample blocks or the distinguishing features of coding blocks.To address these problems,texture synthesis information hiding method combining random mapping and improved image quilting is proposed.Firstly,the sample image is divided into sample blocks and then classified into different categories by sample block means.Secondly,secret information sections are encoded then placed in blank image at particular positions by setting a random mapping relationship between secret information segments and the sample categories.Finally,an improved image quilting method is used to stitch each encoding block with its adjacent blocks in order to generate a stego texture.Experiment shows the proposed method can cover secret information by texture synthesis.Secret information segmentations and encoding sample blocks do not correspond to each other.The proposed method relies entirely on user keyo0uuopppp09oj0l;-s and has some anti-attack capability.
作者
李国利
邵利平
任平安
LI Guo-li;SHAO Li-ping;REN Ping-an(School of Computer Science,Shaanxi Normal University,Xi'an 710119,China)
出处
《计算机技术与发展》
2020年第1期106-111,共6页
Computer Technology and Development
基金
国家自然科学基金(61100239)
陕西省自然科学基金(2011JQ8009,2016JM6065)
中央高校基本科研业务费支持项目(GK201402036,GK201703057)
关键词
纹理合成
信息隐藏
图像缝合
生成式隐藏
样例纹理合成
texture synthesis
information hiding
image quilting
generative hiding
sample texture synthesis
作者简介
李国利(1991-),女,硕士研究生,研究方向为纹理合成信息隐藏;通信作者:邵利平,博士,副教授,研究生导师,CCF会员(11901M),研究方向为数字图像音频置乱、加密、密写、水印、隐匿、分存、伪装和欺骗等;任平安,副教授,研究生导师,研究方向为计算机网络安全。
