Video dual-domain watermarking algorithm based on hybrid DWT-DCT and frame grouping
Aiming to address the compatibility of robustness and reversibility in existing video watermarking algorithms, a video dual-domain watermarking algorithm based on hybrid DWT-DCT and frame grouping was proposed. The algorithm, which was based on the H.264 video coding standard, introduced a video dua...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
POSTS&TELECOM PRESS Co., LTD
2025-02-01
|
| Series: | 网络与信息安全学报 |
| Subjects: | |
| Online Access: | http://www.cjnis.com.cn/thesisDetails#10.11959/j.issn.2096-109x.2025010 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Aiming to address the compatibility of robustness and reversibility in existing video watermarking algorithms, a video dual-domain watermarking algorithm based on hybrid DWT-DCT and frame grouping was proposed. The algorithm, which was based on the H.264 video coding standard, introduced a video dual-domain watermarking framework. Video frames were grouped according to their scene smoothness, and within each group, frames meeting the predefined criteria were selected as hidden frames for watermark embedding. The hidden frames were processed using a hybrid DWT-DCT approach to generate the watermark embedding domain. Robust watermarking was embedded into the low-frequency domain of the Y channel using a robust embedding method, while reversible watermarking was embedded into the high-frequency domain of the Y channel through histogram migration techniques. At the decoding end, watermark extraction and lossless recovery of the original video were achieved. Experimental results demonstrate that the proposed algorithm exhibits excellent invisibility, with the average peak signal-to-noise ratio and average structural similarity index of the embedded frames being 49.009 2 dB and 0.984 0, respectively. The algorithm is shown to be robust against distortion noise attacks and frame attacks of varying intensities. The average normalized correlation between the extracted watermark and the original watermark is above 0.993 3, while the average bit error rate is below 0.058 2. Additionally, the computational complexity is low, with the time difference for unit frame encoding and decoding being less than 0.033 s and 0.025 s, respectively. These results indicates that the proposed algorithm has high application value. |
|---|---|
| ISSN: | 2096-109X |