Saturday, January 18, 2014

JRC Identifies Sensors by Noise

European Commission Joint Research Centre (JRC): Recent advances on image processing allow identification of which device has taken a given picture, and allow a match with those contained in social network accounts (Facebook, Flickr, etc.). This possibility has been demonstrated in a recent JRC study that received the Best Paper Award at the International Conference on Computer Vision Theory and Applications (VISAPP 2014) in Lisbon, Portugal this January:

"On the usage of Sensor Pattern Noise for Picture-to-Identity linking through social network accounts"
Riccardo Satta and Pasquale Stirparo

JRC exploited the noise pattern that is left in a digital picture by any image sensor, as a fingerprint of the imaging device (whether it is a digital camera, a smartphone, a tablet, …). Such a Sensor Pattern Noise (SPN) has been shown to be unique for every single device. This allows one to find, given a picture, a social network account containing images taken by the same camera, and in turn it can be possible to infer the identity of the camera owner.

3 comments:

  1. Several years ago a crime in the Netherlands was resolved by proving that a particular camera (which was in possession of the suspected murderer) was used to shoot pictures stored on a memory card (which was found close to a dead body) ! So the statement that the FPN of a sensor/camera can be seen as its signature is not really new.

    ReplyDelete
  2. For this to work like human fingerprint system, does it mean somebody has to set up and keeps updating an enormous camera database including all existing and to be shipped smartphones, tablets, DSC, etc.? Given the number of cameras to deal with, this sounds a daunting task!

    ReplyDelete
  3. This seems to be quite similar to US7129973B2, filed May 28th 2002.

    ReplyDelete

All comments are moderated to avoid spam and personal attacks.