Secure and robust watermarking using arnold transform and Neyman-Pearson decoder

In this paper, we propose an adaptive multiplicative method to watermark binary data by scaling low frequency wavelet coefficients in the first level of decomposition. We decode message using hypothesis testing with optimum threshold value according to Neyman-Pearson test. We define a detection factor, k0, which is equivalent to Lagrange Multiplier and use PSO optimization algorithm to obtain optimum threshold value. To increase security, we impose Arnold transform with optimum iteration number both on the host image and logo before embedding data. The robustness of the method is analyzed against well-known Stir-Mark attack set and compared with similar methods. Experimental results confirm the imperceptibility of the proposed method and more robustness against attacks.