基于子图像偏差纠正的迭代混合算法

为了提高水印系统的鲁棒性,通过理论分析和实验研究了现有迭代混合算法中混合次数的有限性和鲁棒性问题,指出弱鲁棒性的主要原因在于对误差的累积放大.进而提出了基于子图像间偏差纠正的迭代混合算法.该算法通过修正混合图像的改变来降低攻击带来的影响,同时将水印在密钥控制下进行多份嵌入.新算法的优点为:对于强相关性的攻击的抵御能力强,特别是对于区域性局部攻击,该算法可完全消除攻击对提取水印的影响.结果表明,文中所提出的的改进算法是正确的,其鲁棒性优于原迭代混合算法.

Iterative Blending Algorithm Based on Error Correction of Sub Image

In order to improve the robustness of a watermark system,theoretical analyses and experiments are carried out to investigate the limitation and robustness in blending number of the existing iterative blending algorithm,demonstrating that the main reason for weak robustness is the multiplying of the image error many times.Furthermore,an iterative blending algorithm based on the error correction of sub image is proposed.In this algorithm,the sub images of the watermarked host image are modified by the sub images of the original host image,and several watermark images are embedded by the control of secret keys.The proposed algorithm is of strong robustness in resisting attacks with strong relativity.Especially,to cropping attacks,the error of the extracted watermark is appro-ximately zero,thus concluding that the proposed iterative blending algorithm is more robust than the original one.