Josef Baker Product Manager, Friend MTS
As organised for-profit video piracy continues to present a serious financial threat to content owners and distributors, proven cutting-edge technologies remain essential tools in protecting valuable video content from illicit redistribution. And while live sports content represents the most lucrative spoils for pirates based on the recurring high profit they can get from redistributing it through their widespread networks, all types of video content can be affected by piracy, costing billions in lost revenue each year.
Digital video fingerprinting is a key component in the war on piracy at scale. It is a technology that enables large-scale automated identification of video and therefore comprehensive detection of pirated material across a broad spectrum of distribution channels: social media platforms, mobile apps, streaming sites, pirate IPTV services, illicit streaming devices, add-ons, plugins, etc.
Digital fingerprinting provides content protection and anti-piracy companies with a means of identifying pirated streams among the masses of video material that is captured and analysed by content monitoring systems/platforms. It’s a highly automated process where huge quantities of suspected pirate material need to be processed quickly and accurately. As an illustration, the amount of video that Friend MTS’s automated global monitoring platform captures equates to 8 months’ worth of video per day, every day.
The first step is to produce a fingerprint of the source video. This is a customer’s video that a content security company is tasked with protecting from piracy, and which is provided to it from the original source. Just like human fingerprints are unique to an individual, fingerprints of a video are also unique to each video. They help to automatically identify this video among millions of “suspect” or candidate videos with forensic precision. This level of monitoring at scale is only possible through the implementation of technology that automates the monitoring, capture and recognition process. Whilst it is possible to employ teams of people to manually monitor video (as some monitoring services do), this is of course a very small-scale approach, which not only limits the amount of video that can be processed but introduces the inevitable factor of human error.
Video fingerprinting is a non-intrusive technique for identifying content based on the movement observed between video frames. Fingerprinting technology generates a “snapshot” of the video stream every half second. The video frames are then analysed to determine a unique mathematical signature representing that snapshot. This signature is that video stream’s ‘fingerprint’, a very small data file comprising a series of patterns or characteristics that can automatically identify this video among others.
Once fingerprints of the source video are created, they are stored in a database as reference material. Again, automation and scale matter here: each fingerprint is very lightweight (one hour’s worth of video generates fingerprint data of just 115KB!). Accordingly, these can be massive databases that contain fingerprints of millions of source videos; nevertheless, with the right technology, these reference databases can be easily and quickly scanned.
The purpose of the reference database is to find a match with any candidate videos that are captured by the monitoring systems. Once a candidate video is captured, a fingerprint of it is created. The candidate is then compared to the source video in the database of reference fingerprints. If a match is found, the matching candidate video is confirmed to be a case of unauthorised use of the proprietary video material.
Fingerprinting isn’t the same as digital watermarking – these content protection tools serve different purposes and are based on different technologies. Fingerprinting allows the automatic identification of infringing video material while watermarking pinpoints the source of this unauthorised content.
Fingerprinting relies on an analysis of characteristics within the video itself, so there is no alteration of the original video. With watermarking, on the other hand, additional information is added to the video in the form of watermarks. With modern forensic watermarking technologies, however, the quality of the video remains unaffected, as the watermarks are practically imperceptible to the human eye.
The recent pandemic led to a huge increase in the use of video by businesses across a wide diversity of industries, as a means of information and expertise exchange among geographically distributed offices and across their networks of partners and resellers. The protection of confidential and potentially highly sensitive video information has become a critical factor for many businesses. The consequences of leaking proprietary video content can have significant, highly damaging impacts, ranging from financial losses to reputational and brand damage. Like other content protection technologies, fingerprinting has a huge potential in protecting sensitive content and preventing these negative consequences.
If you want to learn more about fingerprinting or any of our content protection and anti-piracy services, please do not hesitate to speak to one of our experts.