Abstract: The goal of this work is to show an optimization arithmetic on gray-level blind watermarking with biggish resolution embedding in DCT domain based on the owner image．With visual frequency sensitivity,gray scale enhancement,Just Noticeable Distortion（JND） on gray-level image,this arithmetic selects DCT-block to compress watermarking image and make automorphic disorder,then embeds watermarking.This arithmetic has no use for the owner image when extracting watermarking,and embeds a great deal of watermarking.Experimental result proves that it is robust enough to resist JPEG compression,visual image has a higher fidelity,robustness and obstinacy．

摘要： 从改进的角度,提出一种基于宿主图像的DCT域嵌入较大分辨率灰度级数字盲水印的优化算法。该算法采用视觉频率、图像灰度增强和灰度图像门限值JND量化选取DCT子块系数,对嵌入宿主图像的较大分辨率256级灰度水印先行压缩和自同构置乱,再行嵌入宿主图像。该算法在水印提取时无需宿主图像和原始水印的参与,较大幅度地提高了水印的嵌入量。实验结果表明此算法抵抗压缩的能力强,视觉图像无明显降质,具有较强的鲁棒性和顽健性。