Combined DTCWT-SVD-Based Video Watermarking Algorithm Using Finite State Machine

The video carrying the watermark will undergo various attacks such as transcoding, noise, and temporal synchronization during network propagation (upload/ download), and the existing algorithm doesn't involve analyzing and testing the combined attacks encountered by the video during network propagation. In order to correctly extract the water-mark in the video after network propagation to protect the rights of copyright holders, this paper proposes a video watermarking algorithm based on the finite state machine in DTCWT-SVD domain. Firstly, the carrier video is divided into several groups of the frame and the watermark image is scanned into several segments by Zig-Zag. Then, the watermark sequence collection is generated by the finite state machine, and each watermark sequence is embedded into the corresponding video frame by modifying the singular value of the DTCWT low-frequency coefficient matrix. Finally, the watermark is extracted according to the predefined relationship of singular values. The experimental results show that the proposed algorithm can effectively resist transcoding, noise, temporal synchronization or other attacks. It is also can accurately extract watermark information in the video after network propagation, which can effectively resist the combined attacks encountered by video in the process of network propagation.