Robust Watermarking Algorithm Using Gram-Schmidt Based Improved QR Decomposition

Due to the Internet's strong development, the problem of copyright infringement of digital products is increasing. Many solutions have been proposed to protect the copyrights of digital products, of which, robust watermarking is an effective and highly applicable solution. QR decomposition-based image watermarking method not only has positive performance in terms of watermarked image quality and robustness but also conforms to real-time systems. However, when applying QR transformation, some pixel matrices do not ensure the uniqueness of the orthogonal matrix and the invertible upper triangular matrix. This paper proposes a new image watermarking scheme based on QR decomposition with the solution of selecting suitable embedding blocks. Pixel blocks are only applied with QR decomposition and embedded a watermark bit when the matrix determinant is non-zero. This solution helps to limit modifying the pixel values, thereby, increasing the quality of the watermarked image and improving the robustness of the extracted watermark image. The experimental results show that our proposed method provides better quality for the watermarked image and it can effectively extract the watermark image in almost all attacks.