面向多媒体信息内容保护的特征水印技术*

将数字水印技术应用于多媒体内容保护方面，提出了一种面向多媒体信息内容保护的通用算法框架，并给出了其核心技术——内容特征提取和评价的基本准则(即对非内容级操作鲁棒，对内容级操作敏感)。该算法完全基于多媒体信号的内容生成特征水印，进而将其不可见的隐藏于原始信号中，从而达到对原始信号进行内容保护的目的。在对受保护的信号进行内容窜改检测时，只需用同样的方法计算当前内容特征水印，并将其与所提取的嵌入水印相比较即可。由于该算法可以在不需任何额外信息的情况下对各种多媒体信号进行内容保护和窜改检测，因此具有较高的通用性和

A Feature Watermarking Technique for Multimedia Content Protection

Based on conventional digital watermarking technique,a general algorithm framework for multimedia content protection is proposed in this paper.Then,we present the fundamental rule for the extraction and evaluation of content features,the key problem of the framework,that the content feature should be robust to non-content level operations,while sensitive to content level operations.As for the target signal ot be protected,a feature watermark is automatically generated according to its content and is hidden imperceptibly in the original signal.Malicious content level alteration could be detected through comparing the extracted watermark and the feature watermark computed from the examined signal by the same method.Because no additional information is required for the protection and content alteration detection of any kinds of multimedia signal,our solution is more general and advantageous.Experiments on image content protection show it outperforms traditional digital signature based methods.