A digital watermark is a sequence of codes or a signal inserted into digital content in order to determine its ownership and ensure its integrity. Watermarks are also used for source tracking in case of content piracy. Content providers often subject the watermarked videos to several kinds of image processing, such as compression (using MPEG or other compression technology), resizing, filtering, and D/A or A/D conversion.
These image processing techniques are also used by illegal users to retrieve the embedded information. Hence, it is crucial for video watermarking to be robust enough so that watermarks can be reliably detected even after any kind of processing or manipulation. The detection process should also be fast – for example, in the case of live streaming – so that pirated content can immediately be taken down.
If a copy of the watermark is detected later, that is, it is leaked at any point, be it through the OTT platform in the case of DRM protected content, content creator, studio, or pay-TV operator, it can be retrieved from the copy to identify its source. Watermark detection uses the same video watermarking key that is used while inserting the watermark and the reverse embedding method to detect the presence of a watermark or read the information incorporated in it.
A video is an electronic representation of moving images in the form of digital data. The objective of a detector is to associate target objects in consecutive frames of the video. The detector analyzes every frame of the pirated video to identify the original watermark. This technique along with object recognition and artificial intelligence (AI) methods are often used to detect watermarks in videos. Several algorithms have also been developed to detect watermarks that are usually hard to extract such as translucent or diffuse watermarks.
In case the watermark payload is detected, the session database is used to find the relevant session information that matches the payload key value. The detection results can then be shared with the content owner to help them find the exact location of content leakage.