云模型在唱片版权保护中的应用

An album watermark based on Cloud model

提出了一种基于云模型的唱片水印。首先,唱片中每首音乐都通过小波传输得到一个小波系数,以其组合成的巨量的云晶作为零水印;然后利用逆向云生成器计算出云晶的3个特征参数Ex、En和He,利用特征变化率,就可以保护水印并评估原始水印与萃取水印相似度。实验结果表明,该算法能同时有效地保护整张唱片中的所有曲目,还能有效抵抗低滤波、重新采样、压缩以及剪辑等攻击。

This paper proposes an album watermark algorithm based on Cloud model. Firstly, all the single audio in the aggregation are performed by wavelet transform respectively to get wavelet coefficients. The combination of these coefficients are mapped into a considerable number of Cloud drops which are served as 0-watermark. Secondly, It emploies reverse-Cloud generator to calculate the three numerical characteristics E, E, and H, of Cloud drops, utilizes change rate of these characteristics to detect watermark and evaluate the similarity between original watermark and the extracted watermark. Experimental results show that the proposed algorithm can protect copyright of all audio in audio aggregation at the same time, and it is robust to attacks such as lowpass filtering, resampling, requantization, compression, cropping, etc.