|
![]() |
A picture is worth 1 000 wordsDigital forensics for images and videos is growing in importance - the tool can be used to catch criminals and track copyright infringement. EU-funded researchers have developed new techniques to advance the science - this could be a big help for law enforcement and a boost to industry competitiveness.
![]() © leowolfert - fotolia.com Was a video taken with that specific camera? Police investigating a sexual abuse crime would want to know. Has someone tampered with an image to mislead people? A journalist may want to be sure a photo has not been manipulated before using it for an article. Software and methods developed by EU-funded project DIVEFOR can help in such cases – and do much more – using advanced digital techniques. In fact, police in the UK used software refined by the project to prove that an indecent video found on a suspect’s mobile phone had been shot with that same device, and had not been copied or sent to him, as the accused claimed. “We understand that this is the first-ever conviction obtained using this technology, which provides law enforcement with an extra option to help detect offences of any kind in which the source of digital imagery is an issue,” said Sussex Police in a July 2014 statement. The software was originally developed in 2009 by Chang-Tsun Li of the UK’s University of Warwick. Afterwards Li received a Marie Curie fellowship through the EU-funded DIVEFOR project that enabled him to further develop the digital techniques he used for the software. The software is owned by Forensic Pathways, a UK-based company and a DIVEFOR project partner. The software is able to uniquely identify and link images and videos with the devices used to create them. DIVEFOR developed additional forensic applications of the techniques and refined digital identification and authentication techniques. The work improved the original software and produced additional prototype tools related to device identification, licensing infringement and anti-tampering, among others uses. The project partners have received four patents for the work. “The applications for our research based on hidden data in the content also involved authentication, content integrity verification, copyright protection and the detection of covert communications,” says Li. Making sense of pattern noise Light sensors used by all digital video and image devices leave unique traces – known as ‘pattern noise’ – in recorded data. This digital ‘fingerprint’ can be used forensically in many ways, including matching a specific video or image to a particular recording device. DIVEFOR studied the forensic applications of these fingerprints and of other unique recorded data. The researchers also developed secure digital watermarking methods to prevent the tampering of images and videos, and advanced techniques for concealing messages in digital images and videos. In parallel, they developed prototype software to automatically classify a group of images or videos into categories – by device, for example. The software could help police to quickly identify the sources of illegal images, even when they have an enormous amount to process. DIVEFOR ended in May 2014. Since then, the project’s partners – two SMEs and two universities – have been involved in further research to improve the techniques advanced during DIVEFOR. Li will also continue developing DIVEFOR’s research through a new EU-funded project, IDENTITY. “The two participating SMEs broadened the spectrum of their expertise and improved their competitiveness,” Li says. “Five staff members from the two SMEs were also involved in 15 secondments to the two participating universities. They have acquired scientific knowledge in areas that include general research methodologies, multimedia security, digital forensics, image processing and pattern recognition.” Li also points out that in return, six researchers from the two universities involved in the project gained commercialisation and innovation experience through secondments to the SME partners. Project details
See also
|
![]() |
|