A forensic watermark is a set of codes or a signal that is added into digital content to identify who owns it and to assure its integrity. This technology is used to safeguard the interests of its legitimate users and owners against piracy attempts. Watermarked videos are frequently subjected to image processing, such as compression (using MPEG or other compression techniques), resizing, filtering, and D/A or A/D conversion by content producers during the encoding process.
Pirates also operate in the same technological territory. They also employ the image processing techniques on DRM protected content to extract the embedded information so that it can be fully removed or distorted.To defeat these efforts, video watermarks must be sufficiently resilient such that they can be reliably detected even after extensive processing or alteration attacks carried out by smart hackers. The identification/extraction method should also be quick – as in the case of live streaming – so that stolen content may be removed quickly from illegal streams.
Even if a copy of the watermark video file is discovered later by the creator’s intelligence network, its watermark can be extracted from the copy to determine the source of leakage. The watermark-extraction process identifies the point at which the video asset was leaked, be it in the OTT platform’s distribution chain, at the content developer’s table, within studio, or through a pay-TV provider. To identify the presence of a watermark or read the information embedded in it, the reverse embedding approach is utilized in conjunction with the same video watermarking key that is used while embedding metadata.
A video is a digitally encoded electronic representation of moving visuals called frames. A detector’s goal is to associate target items in video frames that are consecutive. To identify the original data embedded during the video watermarking process, the detector examines every frame of the video it feels could be pirated. Watermarks in videos assets are frequently detected using this technique as well as object recognition and artificial intelligence methods.The multi-DRM vendor whose services are used to embed watermarks have also developed algorithms for detecting watermarks that are otherwise difficult to extract, such as translucent or diffuse watermarks.
If the detector finds the watermark payload in the target video, it is further examined for the session information and matched against entries in the watermark database maintained by the multi-DRM vendor. One by one, all metadata values of the watermark are matched against the database values. The results of the detection and matching process is then provided to the content owner to assist them in locating the exact location of the leakage.
