[[MPEG-7]] is not three more generations down the road of the hugely popular MPEG-4 video compression standard.
It is a video content description technology that is used to helps users search through vast quantities of video content faster than ever before possible. Finally, a machine-relevant meta tag for the video industry. This is a big deal, because broadcast and video entertainment are such large industries. They need this technology to automate many of their background processes and business processes, including piracy inspections.
I did a piracy inspection for Brockmann & Company the other day. I searched for Brockmann and .pdf and discovered a handful of AV resellers, republishing our content without a license. I wrote to them reminding them of my Copyright (one was in the UK) and asked them to acquire a license. Instead they apologized and took down the offending link. Shame on them.
The Economist print edition of May 15, 2010 wrote about how MPEG-7’s techniques could compare the small video signature of a movie to the zillions of bytes of movie clips data stored in YouTube’s vast video store. Even doctored or recorded from a cinema movies and clips can be identified.
NEC and Mitsubishi Electric’s methods compare the brightness and assign a -1, 0 or +1 to each of 380 predefined regions of interest in a single frame of the video in question. This could be done for some or all of the frames in a movie. It’s apparently immune to doctoring, poor quality or a different hue since there are so many regions and since the information about relative luminescence is retained despite the sharpness of the images.
This technique generates a 96% average detection rate, with very low false positives – 5 per million.
