基于神经网络分类的图像数字水印算法

AN ALGORITHM OF IMAGE DIGITAL WATERMARKING BASED ON NEURAL NETWORK CLASSIFYING

数字水印技术是一个新的多媒体安全研究领域,它区别于传统的加解密系统,它的优点在于容易实现,适合给信息量庞大的多媒体数据提供如版权保护、所有权认证、票据电子防伪等安全措施。本文讨论了常见的商品化数字水印算法的鲁棒性,并将自组织特征映射(SOFM)算法应用到数字图像水印算法中,此算法先将图像分类,然后对不同的类用混沌序列产生不同强度的水印信号,将水印信号叠加到原始图像中,水印检测算法将不同子块的信号相关性加权,若加权后的相关性大于指定的阈值,则说明存在水印,实验结果表明,此水印算法具有较高的鲁棒性。

Digital watermarking technology is a new research field of multimedia security. It is different from traditional encrypt/decrypt system. It has the advantages of easy implementation and capability of providing wide security services such as copyright protection, authentication and secret communication. A new digital watermarking algorithm is proposed in this paper based on analyzing some existing algorithms. Self-Organization Feature Map (SOFM) algorithm is applied to digital image watermarking. In the proposed algorithm, the image is classified into some sub-classes through SOFM, then different strength watermark generated by different chaotic sequence is added in DCT domain according to different frequency of each sub-class. Watermark detection algorithm weights the correlative coefficient of different sub-classes. If the weighted value exceeds to the given limit, then there exists a watermark in the image. Experiment results show this algorithm is strongly robust.