The aim of watermarking is to embed subliminal information in content files in order to ensure protection against piracy and identify the content creator/distributor’s license to the user. The watermark essentially contains a digital signature which is exclusively limited to the content creator. The watermark extraction procedure, thus, only permits the owner or a few authorized persons with access to the secret keys to retrieve the watermark from the video file, thereby preventing unauthorized usage of DRM protected content.
In case of an infringement, watermarks need to be extracted to identify the exact source of the leakage. Based on the requirements for watermark extraction and detection, algorithms can be classified into non-blind, semi-blind, and blind techniques.
Non-blind video watermarking schemes need the original content copy, watermark, and the secret key for the content validation process and extraction. Non-blind watermarks record the traffic pattern of incoming flows and correlate them with outgoing flows. They also aid traffic analysis by applying some modifications to the communication patterns of the intercepted flows. The practicality of the non-blind watermarking scheme is limited as it needs extra storage to maintain the source content and watermark data.
Semi-blind approaches (also called “informed” watermarking schemes), on the other hand, require the watermark and secret key. The operator mark is often used to identify the relevant database for the watermarking information such as the session, embedding parameters and possible content information in case of semi-blind and non-blind approaches.
Blind watermarking schemes only need the secret key, which is generally used to generate a random sequence during the watermark embedding process. In blind watermarking, the watermark sequence is directly correlated with the coefficients of the receivent content file and the correlation coefficient is then compared with some detection threshold. Hence, only the watermark sequence and the scaling factor are needed for watermark detection.
In case of blind attempts, the detector must automatically account for the temporal and spatial issues as part of its functionality. Since the original content is not required, blind approaches are the easiest to deploy and are often used when the acquisition of the source content is difficult. However, non-blind extraction techniques may enhance the robustness and security of the solution since the attacker will not have access to the original, non-watermarked content.
