Abstract: Based on the human visual system, a double image watermarking algorithm in wavelet domain was proposed. A piece of meaningful binary image was taken as watermark. On the basis of the original image analysis, the watermark was appended to appropriately selected parts. And double embedment of the watermark was performed through modifying low-frequency and high-frequency coefficients of the selected parts. It is shown that the image embedded watermark has better transparency, and two watermarks robustness complements each other. The algorithm is more robust for handling general images with JPEG compression and noise images.

Key words: Digital watermark, Human visual system, Wavelet, Complementation of robustness 

摘要： 基于人类视觉模型，提出了一种双水印算法。选取一幅有实际意义的二值图像作为水印，在对原始图像进行分析的基础上，选取适宜的图像块加入水印。分别通过修改相应图像块的低频系数和高频系数来进行水印的双重嵌入。实验证明嵌入水印后的图像具有更好的隐蔽性，实现了两个水印的鲁棒性互补。并且对JPEG压缩、图像添加噪声等常规图像处理操作具有较强的鲁棒性。

关键词: 数字图像水印, 人类视觉系统, 提升小波变换, 鲁棒性互补