摘要
在目前基于分块的无损图像认证算法中,在图像分块中嵌入水印需要合并较大的图像区域,这将导致窜改定位的精度降低。为提高认证精确性,提出基于三叉树划分的认证算法。该算法预先串联对图像三等分后产生的基本块,形成基本块链表;然后选定适当大小的图像块作为认证块进行水印嵌入,当水印无法嵌入时,从基本块链表中选取一个基本块进行认证块扩展。基本块的细粒度特性降低了为嵌入水印信息而对图像认证精度造成的影响。在对认证信息进行签名时,通过重复嵌入无损水印的方式来提高算法的鲁棒性,防止矢量量化攻击。实验结果表明该算法窜改定位能力优于基于自适应图像块组合的无损图像认证算法。
As the existing block-wise lossless image authentication algorithms will sacrifice the precision of locating distortions,this paper proposed the lossless image authentication algorithm based on trisected image basic blocks. The algorithm trisected the original image recursively and connected all the basic blocks into a link. An authentication block was made up of several basic blocks. When the watermark couldn't be embedded into the authentication block,the algorithm tried to merge only one basic block from the link into current authentication block. As a result,it expanded each authentication block's size at a minimum level. Moreover,in order to prevent vector quantization attack,it took a digital signature algorithm and repeatedly embedded the signature into the authentication block via a lossless watermarking algorithm. The experimental results demonstrate that the algorithm is more accuracy than the algorithm based on adaptive combinations of image basic blocks.
出处
《计算机应用研究》
CSCD
北大核心
2014年第7期2224-2227,2232,共5页
Application Research of Computers
基金
广东省高等职业教育教学改革立项项目(20120202029)
广东农工商职业技术学院院级课题(XYYB1107)
关键词
无损图像认证
可逆图像认证
三叉划分图像认证块
矢量量化攻击
lossless image authentication
reversible image authentication
trisected image basic blocks
vector quantization attack