Project: Forensic Image Identifier and Analyzer

The Forensic Image Identifier & Analyzer is an innovative proposed software service for visual imagery that helps police, security and other agencies automatically identify, analyze and link images on web sites, seized computers and storage devices._x000D__x000D_The service is based on technologies developed by two European SMEs, Videntifier Technologies in Iceland, and Forensic Pathways in United Kingdom, with research input from INRIA in France. The goal of this Eurostars project is to develop this service by incorporating the technological advantages of both companies – Videntifier Forensic’s high-capacity multimedia content identification and Forensic Pathways’innovative signal-processing based camera identification. _x000D__x000D_When police arrest a suspect in connection with online crime (distribution of child abuse material, terrorist or hatred propaganda, large-scale copyright violations), the police often seize the suspect’s digital devices for evidence gathering.. However, finding the image based evidence, identifying it, crosslinking, and cataloguing is time-consuming and tedious. Videos and images files contain a massive amount of data and information, and can't be searched through as easily as text. The problem is identifying and separating suspect material from non-suspect material as well as proving where the material came from. Being able to state that particular material came from a particular camera or that suspect images in a storage device like a laptop are related to each other in terms of their source is crucial._x000D__x000D_Commercially available forensic software offers very little support for investigating multimedia material. Therefore police investigators may end up either manually screening whole collections – or only taking a few samples. In addition, the volume of data found on a single computer has multiplied, leaving the investigator manually screening irrelevant media data for days if not weeks. Much material is never actually analysed in investigations, resulting in serious offenders escaping justice. _x000D__x000D_Key questions to be answered are often “Did this image or video originate from this device?” Or, “Is this image or video related to any of the other video or images in this database?” “Does this image or video contain the same content as any other image or video in the database?” Linking together related evidence material and cases is also important as well as identifying child abuse cases, finding the victims themselves or obtaining additional evidence of abuse. _x000D_Videntifier Technologies, a high-tech company in Iceland has been working on image content identification since being founded in late 2007. The company has developed an automated service, Videntifier Forensic, that assists police authorities by automatically identifying video files on seized storage devices by looking at the visual content within._x000D__x000D_Forensic Pathways, a company from Staffordshire, UK, has developed a technology for analyzing the sensor-pattern noise of cameras within digital images to automatically identify both the camera model and individual cameras that created an image. The technology can group a large set of images by the camera that took them, and like Videntifier it relies solely on processing visual information._x000D__x000D_The benefits of Videntifier technology is identification of content material in the visual image or video and the benefit of Forensic Pathways technology is the identification of a digital signature left by the silicon chip in the image._x000D__x000D_Both SMEs have understood that their technologies are complementary and a joint product would do even better on the market. Collaborative effort will also allow for a much more efficient sales process encompassing a ‘One Stop Approach to Image and Video Identification, Classification and Attribution.’ The goal of this Eurostars project is to develop and market the joint product, namely the Forensic Image Identifier & Analyzer. _x000D__x000D_The key milestones within the project are these:_x000D_1. Extend Videntifier Forensic's video identification to still images._x000D_2. Extending the image identification to identifying still images within videos._x000D_3. Extend the camera identification to add the capacity to identify video camcorders._x000D_3. Build a central database with the sensor signature of the 1000 most commonly available camera and camcorder types._x000D_4. Speed up the sensor fingerprints extraction and camera classification process by migrating the algorithms onto Graphics Processing Units (GPU). _x000D_5. Integration of the two sets of technology into a product._x000D_6. Extending the service functionality to identification of material on web-sites._x000D__x000D_Solving scalability issues will be important. Therefore the TexMex lab of INRIA in Rennes, France joins the team. They have significant research experience on content-based image search at the vast millions-of-images scale and also has access to hundreds of millions of research images, especially to evaluate search performance and robustness on such a such a huge scale.

Acronym FIIA (Reference Number: 6687)
Duration 01/03/2011 - 20/12/2015
Project Topic FIIA is an innovative software service for the Forensic market that automatically identifies and analyzes the content of images on web sites and seized computers. The service saves time and money, gathers better evidence, and builds stronger court cases.
Network Eurostars
Call Eurostars Cut-Off 6

Project partner

Number Name Role Country
3 INRIA TexMex Project-Team Partner France
3 Forensic Pathways Partner United Kingdom
3 Videntifier Technologies ehf Coordinator Iceland