Improved High Capacity Spread Spectrum-Based Audio Watermarking by Hadamard Matrices

摘要

Traditional spread spectrum-based audio watermarking methods usually use randomly generated pseudonoise sequences for watermark embedding and extraction. In this paper, we use Hadamard sequences, which are rows of Hadamard matrices, to embed and extract watermarks instead of pseudonoise sequences. By exploiting the orthogonality of Hadamard sequences and a technique of sign change, we propose a new spread spectrum-based audio watermarking method. Experimental results show that, compared to the newly high embedding capacity spread spectrum-based audio watermarking method, our method achieves a better perceptual quality and a higher embedding capacity while maintaining almost equal strong robustness. We also provide a theoretical analysis of the security of our method.