Blind watermarking schemes have broader applied prospects in copyright protection of digital products, because it doesn' t need the host image in watermark detection procedure. The proposed blind watermarking algorithm uses adaptive selected embedding policies of relationship embedding and quantification embedding on the basis of blocking discrete cosine transform (DCT). The relationship refers to the magnitude relation between the DC coefficient of 8 × 8 image block DCT and the second low-frequency coefficient lined in zigzag pattern. The quantification embedding uses jittering modulation and can easily adjust the watermark transparency and robustness. A great deal of simulation experimental results show that the algorithm has fairly good robustness against the attacks including noise, filtering, JPEG compression and cropping,etc. ,as well as maintains satisfying transparency.%盲水印提取时不需要参考原始图像,在数字产品的版权保护中具有更广泛的应用前景.提出的盲水印算法在分块DCT的基础上,采用了关系嵌入和量化嵌入自适应选择的嵌入策略.关系为8×8图像块DCT的直流系数和之字形排列的第二个低频系数之间的大小关系.量化嵌入采用抖动调制,可以方便地调节水印的透明性和鲁棒性.大量仿真实验结果表明它对噪声、滤波、JPEG压缩、剪切等攻击有很好的鲁棒性且具有良好的透明性.
展开▼
