Bibliographic record
Abstract
The growing problem of the unauthorized reproduction of digital multimedia data such as movies, television broadcasts, and similar digital products has triggered worldwide efforts to identify and protect multimedia contents. Digital watermarking technology provides law enforcement officials with a forensic tool for tracing and catching pirates. Watermarking refers to the process of adding a structure called a watermark to an original data object, which includes digital images, video, audio, maps, text messages, and 3D graphics. Such a watermark can be used for several purposes including copyright protection, fingerprinting, copy protection, broadcast monitoring, data authentication, indexing, and medical safety. The proposed thesis addresses the problem of multimedia protection and consists of three parts. In the first part, we propose new image watermarking algorithms that are robust against a wide range of intentional and geometric attacks, flexible in data embedding, and computationally fast. The core idea behind our proposed watermarking schemes is to use transforms that have different properties which can effectively match various aspects of the signal's frequencies. We embed the watermark many times in all the frequencies to provide better robustness against attacks and increase the difficulty of destroying the watermark. The second part of the thesis is devoted to a joint exploitation of the geometry and topology of 3D objects and its subsequent application to 3D watermarking. The key idea consists of capturing the geometric structure of a 3D mesh in the spectral domain by computing the eigen-decomposition of the mesh Laplacian matrix. We also use the fact that the global shape features of a 3D model may be reconstructed using small low-frequency spectral coefficients. The eigen-analysis of the mesh Laplacian matrix is, however, prohibitively expensive. To lift this limitation, we first partition the 3D mesh into smaller 3D sub-meshes, and then we repeat the watermark embedding process as much as possible in the spectral coefficients of the compressed 3D sub-meshes. The visual error of the watermarked 3D model is evaluated by computing a nonlinear visual error metric between the original 3D model and the watermarked model obtained by our proposed algorithm. The third part of the thesis is devoted to video watermarking. We propose robust, hybrid scene-based MPEG video watermarking techniques based on a high-order tensor singular value decomposition of the video image sequences. The key idea behind our approaches is to use the scene change analysis to embed the watermark repeatedly in a fixed number of the intra-frames. These intra-frames are represented as 3D tensors with two dimensions in space and one dimension in time. We embed the watermark information in the singular values of these high-order tensors, which have good stability and represent the video properties. Illustration of numerical experiments with synthetic and real data are provided to demonstrate the potential and the much improved performance of the proposed algorithms in multimedia watermarking.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".