在充分考虑图像视觉感知特性的基础上,提出一种图像自适应半脆弱水印算法。首先利用图像亮度、纹理特性确定水印嵌入比特深度;然后采用自适应LSB替换方法嵌入水印信息,水印嵌入强度依赖于载体图像特征;最后应用优化像素调整进一步提高水印透明性。实验结果表明,算法对偶然攻击有一定的鲁棒性而对恶意攻击比较敏感,且能有效识别图像攻击类型。此外,理论分析与仿真实验均说明本文算法比同类算法具有更高的透明性,其隐秘图像删值要高约2.90dB。
An image adaptive semi-fragile watermarking algorithm was proposed by considering the human visual perception. First, the bit depth to be embedded was determined by exploiting the luminance, and texture masking characteristics. Then, the watermark is embedded into the host image with adaptive least significant Bit substitution. The watermark strength is dependent on the host image features due to the use of the visual perception features. Finally, an optimal pixel adjustment process is used to further enhance the transparency. Experiment results show that the proposed algorithm is sensitive to malicious attacks while robust to incidental attacks, and can classify effectively the type of attacks encountered on the watermarked image. When compared with some existing semi-fragile watermarking methods, the proposed scheme achieves about 2.90 dB higher PSNR