【摘要】 数字水印是一种有效的著作权保护的方法。提出了一种新的水印信息失真估计和神经网络融合的盲水印算法。在使用一种特殊的数据规范化构建方案来评估提取水印证据的失真度之后,运用神经网络提取水印算法,还可以进行剪切定位。该水印方案也不需要原始图像,并不需要增加额外的水印容量。最后实验证明,该水印算法在保持不可觉察性外水印的鲁棒性方面更强。更多还原

【Abstract】 Digital watermark is an effective copyright protection method.A new robust public watermarking algorithm is proposed with the base of the method of confidence evaluation and neural network data fusion.First a special addition formula is applied to later evaluate the confidence of each extracted watermarked bit in the extraction.Then a normalization is used to accord with the approximate distortion rate.Moreover the neural network is applied to extract the watermark.The algorithm doesn’t need the original image during the extraction.The new watermark confidence evaluation relatively reduces the volume of the embedding watermark compared with the normal method.Finally the experiments show that the watermarking algorithm is imperceptive and more robust.更多还原