Popular Posts

Tuesday 29 November 2011

Altered-image ratings tell you just how fake photos are

Airbrushed images of models and celebrities have already been linked with eating disorders and body dissatisfaction. Now such images could come with a warning, thanks to a system for rating degrees of photograph manipulation.

Politicians in countries such as the UK, France and Norway have already called for labels on retouched images, but the publishing industry has so far resisted the change since nearly all photos are tweaked in some way.

"It's a blunt instrument because you don't distinguish between white balancing or cropping the image and reducing the body by size by 20 per cent or airbrushing every wrinkle to oblivion," says Hany Farid, a researcher in digital image forensics at Dartmouth College in Hanover, New Hampshire. That is why Farid and his colleague Eric Kee have come up with a system that automatically rates retouching on a scale of 1 to 5, from minor changes to a complete digital makeover – you can see see before and after examples on his website.

Farid calculates this rating based on eight statistics that summarise changes in shape, colour and texture. Four of these describe the movement of pixels in the photo subject's face and body, while the others relate to the amount of blurring, sharpening or colour correction in the image. He then used Amazon's Mechanical Turk service to turn these into a single meaningful rating by asking 390 people to assess pairs of images on a 1 to 5 scale, to map a connection between the eight statistics and the human perception of photo manipulation.

The automated system agreed with the human ratings 80 per cent of the time, disagreeing for images where alteration of a few pixels led to a major perceptual change, such as an image of a man with missing teeth that were restored by the photo retoucher. Farid says that further training on a wider variety of images would help improve the system and it could also be modified to work on any kind of picture, not just ones containing people.

Labels generated by the system could be published alongside modified images, but Farid is also working with Kevin Connor, a former product manager at Adobe, to create a plug-in for Photoshop that would rate images in real-time during the editing process, warning retouchers if they stray too far from reality.

"I think that's certainly got some merit," says British MP, Jo Swinson, who has campaigned against retouched images. "Anything that would encourage people to think twice and not automatically go ahead with a lot of retouching just because it's possible could be helpful." But she warns that simply labelling images may not be helpful if the ratings are too small or if their meaning is unclear.

Journal reference: Proceedings of the National Academy of Sciences, DOI: 10.1073/pnas.1110747108

If you would like to reuse any content from New Scientist, either in print or online, please contact the syndication department first for permission. New Scientist does not own rights to photos, but there are a variety of licensing options available for use of articles and graphics we own the copyright to.


Only subscribers may leave comments on this article. Please log in.


Only personal subscribers may leave comments on this article


Subscribe now to comment.


All comments should respect the New Scientist House Rules. If you think a particular comment breaks these rules then please use the "Report" link in that comment to report it to us.


If you are having a technical problem posting a comment, please contact technical support.

Related Posts Plugin for WordPress, Blogger...