.jpg)
British expert.Hugh Carr-Archer, whose British biometrics firm Aurora produces software that identified rioters in a city in the North-West and is used by the Passport Service at Heathrow Airport, said images from social media had their limits but were often a crucial timesaver for police in crisis situations.
"A lot of security forces are relying upon evidence from the public and the public will give it and share it," he told The Telegraph.
The pixellated close ups and a video posted by the FBI onto Youtube may herald an age of more open collaboration between investigators and the public, where every tweeted image and smartphone video needs to be sifted through to find perpetrators quickly.
"The more evidence you've got the greater chance you've got of finding them. It's beneficial. The problem of course is that the quality can be really quite low, for example with the face, but we've found that modern digital CCTV your picture isn't pixelllated at all, so you're more likely to get matches," Mr Carr-Archer added.
"If you've got jerky images from social media, you may have more info but it may not be of more use," he conceded.The CEO of Northampton-based Aurora described the process the FBI may have undertaken in the last few days to the Telegraph as more of a "tool to cut down time" than an exact science, but one that will have saved police precious minutes in the aftermath of the bombings on Wednesday.
"The basic thing is that facial recognition is no different from finger printing; it all works in the same way," he said.
It would have been "very difficult" to identify the suspect wearing dark glasses in the FBI image, as eyes are crucial for matching a profile to a face.
"If the frame is to the side no one in the world would be able to tell you who they are," he said. "You need both eye points."
Profiles of suspects are built up from multiple images of a face at different angles and then "overlayed" onto official shots from passport and driving licence databases.
The software is now so advanced that it can "rank" faces in databases by their likely match to the image of comparison.
The fact the FBI released the images and appealed to the public for help may indicate it lacked a good match from such databases, which can be flawed in sometimes lacking up-to-date pictures.
"People's faces change a lot," Mr Carr-Archer said, especially between early adolescence and the mid-twenties. If one of the suspects had applied for a driving licence aged 15 (as is legal in the US), his face may have changed considerably from the image portrayed on it, he said. "If you have a ten-year-old photo on a passport, you can look very different."
Where facial recognition software comes into its own is in its ability to homogenise images taken at different distances into one "template" of a face, where everything from below the forehead and above the chin is digitised and compared against other images of the same individual. A similar angle is therefore more of an asset than shots taken at an identical distance.
Not every British police force is using such innovation to their advantage. "They will have people sitting there with thousands of hours of footage and when they see a frame that's interesting they'll pause it.
"But a human being can only concentrate really intently for 20 minutes. After an hour they will start to miss things they would have spotted earlier," Mr Carr-Archer noted.
As an article in the Massachsuetts Institute of Technology (MIT) Technology Review said yesterday, "Deploying facial recognition software in the Boston investigation isn’t straightforward because the images available are very different from the evenly lit, frontal, passport-style photos stored in law enforcement databases."
But such software can work faster than a human being could hope for, provided it is fed the right type of image.
However, we are a "significant way off" being able to overlay a face from a passport onto hours of footage from a crowd of people, Mr Carr-Archer said.
