Self-Adaptive Video Watermarking Based on the Motion Characteristic Detection and the Model of Entropy

A new self-adaptive video watermarking, which is combined the motion characteristics detection with the model of the entropy is proposed. By uniting the human visual masking system (HVS) with the blocking-matching techniques we acquire the motion information of every macro-block in video sequence such as the velocity, the motion angle, and the deformation. The motion information is statistic by using the motion entropy. According to the entropy of motion information in each macro-block and the human visual masking system, a set of non-linear formulas have been used to calculate the maximum of the embedding strength of the every block in the video sequence to improve the imperceptibility of the watermarking. The results of imperceptible experiments indicate that the imperceptibility of the video watermarking has been improved by introducing the motion characteristics detection and the model of motion entropy. In the robust experiments the proposed algorithm could resisted the normal video watermarking attracts and has the excellent robustness and security.