Adaptive Motion Compensation and Quantized Embedded Watermarking Against Collusion Attacks
Hongwei Niu, Lixuan Zhu · 2023
Digital watermarking was an important method for video copyright protection, but it may face inter-frame linear attack in two kinds of collusion scenarios. By constructing collusion model, simulating attack principle and analyzing noise source, a hybrid resistance strategy and evaluation criterion based on adaptive motion compensation and adaptive quantization embedding were proposed. The experimental results show that the adaptive motion compensation algorithm has stronger convergence and modulating inter-frame redundancy than pixel and block motion compensation algorithms, the average reduction of PSNR is 28.76%, and the average amplitude of PSNR is 5.67%, which is significantly better than that of the contrast method, which is 46.45%, 37.36%, and 16.33%, 10.81%, respectively. The Peak signal-to-noise ratio of PSNR is higher than that of the contrast method, it is shown that the proposed method balances robustness and concealment while improving the capacity of embedding. In conclusion, the proposed method meets the evaluation criteria and is superior to the comparison method, which ensures that the collusion rules can not be obtained by using data redundancy and the collusion rules can implemented by the collusion method.