Synchronization-enhanced dual-watermarking for video via adaptive dual-tree complex wavelet transform and Zernike moments

Zernike moments offer strong robustness against geometric attacks in image watermarking, but their high computational complexity prevents direct application to video. Additionally, existing Zernike-based video watermarking schemes are often vulnerable to frame order modification attacks. To address these issues, we propose a robust dual watermarking scheme for video based on Dual-Tree Complex Wavelet Transform (DTCWT) and Zernike moments. Specifically, we introduce the DTCWT to adaptively obtain the low-frequency subband of video frame watermark embedding units. Zernike moments are calculated on these subbands, and the copyright watermark is embedded into them. This adaptive DTCWT approach reduces the area for Zernike moment calculation, lowering computational complexity and enabling applicability across diverse video resolutions. In addition to embedding copyright watermarks in video frames, this scheme also embeds synchronization watermarks in video frames to reduce the impact of video frame order attacks. The synchronization watermark corrects the frame sequence, while the copyright watermark verifies ownership. Experimental results demonstrate that, compared with the existing state-of-the-art video watermarking schemes based on Zernike moments and other non-Zernike moments, our proposed scheme shows optimal robustness against geometric attacks, non-geometric attacks, temporal synchronization attacks, and combination attacks.