University: University of St Andrews
Sector(s): Creative Industries, Information & Communications Technologies
Current methods of automatic video data copying (over the Internet and other media) detection that use file names can easily be fooled by simply changing the name. Embedding information into the videos (watermarking) relies on the video not changing significantly when copied, but videos that are copied across peer-to-peer networks can be substantially altered. For example changes in spatial resolution, CODECS, colour depth and frame rates all increase the difficulty of retrieving a watermark for identification. This kind of transformation is commonplace for video which is being distributed via peer-to-peer networks in order to reduce download size or easily fit onto specific type of media (CD, DVD etc). There are also a great number of originals which have not been watermarked. The advantages of this invention over the problems of existing technology are summarised in Key Benefits below.
- Automatically identifies computer derived/digital files containing copies of videos and movies.
- The performance is independent from CODEC, low bitrate, low frame rate, and noise artifacts.
- Does not embed additional information in the video.
- This reliable automatic detection system will assist companies in making cost savings by identifying, confirming and tracking non-authentic video film exchange over the Internet.
- The Motion Picture Association of America (MPAA) estimate that the predicted losses through copyright infringement to companies in the video film distribution marketplace were US$3bn in 2003/04 world-wide and this loss is predicted to grow to an estimated level of US$7bn in 2004/05. The total market size for video film distribution globally is US$179bn.
Demonstration software has been developed and is undergoing refinement. The University has granted patents in the USA, Canada, and Europe (see US 8,433,108). The research group continues to perform Research and Development in the software security area. There have been no commercial parties involved in the research and the University would welcome enquiries from commercial parties interested in developing and marketing commercial applications of this digital fingerprinting technology.