Generalized patchwork algorithm for image watermarking

Abstract.

This paper presents the generalized patchwork algorithm (GPA), which is an extension of the modified patchwork algorithm (MPA). The MPA embeds watermarks additively and changes the mean of the pixels accordingly. In this paper, we take multiplicative watermarks into consideration where the GPA can change the variance of pixels. The GPA combines both additive watermarks and multiplicative watermarks. The embedding functions of the GPA determine embedding parameters adaptively according to the host signals. Detection functions also determine thresholds adaptively. Embedding and detection functions based on the test of statistical hypotheses are presented in this paper. Simulation results show the effectiveness of the GPA. The proposed algorithm is sufficiently robust against various signal processing operations, especially against compression attacks.