A Novel Watermarking Method using Hadamard Matrix Quantization

One of the most used watermarking algorithms is Singular Value Decomposition (SVD), which has a balanced level of imperceptibility and robustness. However, SVD uses a singular matrix for embedding and two orthogonal matrices for reconstruction, which is inefficient. In this paper, a Hadamard matrix is used to get a singular matrix for the reconstruction process. Moreover, SVD works with a floating-point value, which takes long processing time, while the Hadamard matrix works with an integer range, which is more efficient. Visual measurement showed that SVD and the new method had average NC values of 0.8321 and 0.8293, whereas the average SSIM values resulted in the same value (0.9925). In terms of processing time, the proposed method ran faster than SVD with an embedding and extraction time of 0.6308 and 0.2163 seconds against 0.8419 and 0.2935 seconds. The proposed method successfully reduced the running time while maintaining imperceptibility and robustness.