基于心理声学模型的自适应量化同步数字音频水印算法

针对传统的固定步长量化方法对不同种类音频载体适应性较差的缺点,基于心理声学模型,在获取的语音信号的DCT域的低频系数中量化地嵌入水印信息,量化步长由掩蔽阈值自适应控制,以满足水印的盲检测。同时引入了同步信号的思想,利用同步信号和水印序列号定位水印的嵌入位置。实验结果表明,该算法不仅具有较好的不可感知性,而且对诸如低通滤波、有损压缩、重采样和剪裁等攻击均具有较好的顽健性。

Adaptive quantization and synchronous digital audio watermarking algorithm based on psychoacoustic model

The existing quantization audio watermarking algorithms have generally adopted the fixed quantizations strategy, which is unavailable for variety kinds of audio. Watermarking was embedded by quantizing DCT low-band coefficients of digital audio signal. The embedding strength was adaptively controlled by the masking threshold. The idea of embedded synchronic signal is proposed, which uses the synchronic signal for locating the watermark location. Experiment results show that the propsed watermarking scheme is inaudible and robust against various signal processing, such as lowpass filtering , noise adding, re-sampling and cutting.