DrewWilson
April 14th, 2009, 08:20 AM
AN ORWELLIAN nightmare it may be to many of us, but CCTV is a boat full of holes to the organisations that pay for it. That's because the people watching CCTV images back in the control rooms often have too many screens to monitor at once, and so may miss the criminal or antisocial activities they are there to spot.
To the rescue of Big Brother's limited attention capabilities come Ulas Vural and Yusuf Akgul of the Gebze Institute of Technology in Turkey, who have developed a gaze-tracking camera system that watches the eyeballs of CCTV operators as they work. It then automatically produces a summary of the CCTV video sequences they have missed during their shift. "This increases the reliability of the surveillance system by giving a second chance to the operator," the researchers write in the journal Pattern Recognition Letters (DOI: 10.1016/j.patrec.2009.03.002).
The system uses webcam-style cameras trained on the irises of the CCTV operators. From this, software works out where the operators are looking as they stare at each monitor - and the areas they have not been paying attention to. From this it creates a video of what they missed, for them and their bosses to watch at the end of their shift.
To make sure the summary can be watched as quickly as possible, Vural and Akgul have developed an algorithm that discards frames that show only the background with no people or moving vehicles in them, to leave only a few key frames for each scene of interest. Vural says the system runs on a standard PC and processes the images in real time, so the summary frames are ready to browse, like a fast-motion flip book, at the end of the shift.
More... (http://www.newscientist.com/article/mg20227035.500-eyeball-spy-turns-the-tables-on-big-brother.html)
...
To the rescue of Big Brother's limited attention capabilities come Ulas Vural and Yusuf Akgul of the Gebze Institute of Technology in Turkey, who have developed a gaze-tracking camera system that watches the eyeballs of CCTV operators as they work. It then automatically produces a summary of the CCTV video sequences they have missed during their shift. "This increases the reliability of the surveillance system by giving a second chance to the operator," the researchers write in the journal Pattern Recognition Letters (DOI: 10.1016/j.patrec.2009.03.002).
The system uses webcam-style cameras trained on the irises of the CCTV operators. From this, software works out where the operators are looking as they stare at each monitor - and the areas they have not been paying attention to. From this it creates a video of what they missed, for them and their bosses to watch at the end of their shift.
To make sure the summary can be watched as quickly as possible, Vural and Akgul have developed an algorithm that discards frames that show only the background with no people or moving vehicles in them, to leave only a few key frames for each scene of interest. Vural says the system runs on a standard PC and processes the images in real time, so the summary frames are ready to browse, like a fast-motion flip book, at the end of the shift.
More... (http://www.newscientist.com/article/mg20227035.500-eyeball-spy-turns-the-tables-on-big-brother.html)
...