IT Forensics
Methods and software for detecting illegal multimedia data
Investigators today are severely overburdened with the IT forensic analysis of data storage devices and the search for illegal content, such as child pornography. The search for illegal content is often like looking for a proverbial needle in a haystack. While there are solutions for the automated search of data sets and the identification of illegal content, these only work in investigations if the perpetrators have not taken countermeasures.
Example: Image Search: Currently, child pornography is automatically detected in data collections only if it is already known to investigators and exists as an exact copy in a corresponding database. Such material is then matched to the known material using a cryptographic hash value. The weakness of this approach is that even minimal changes to the data alter the hash, making identification impossible. This can occur, for example, when converting to a different file format or, in some cases, simply by opening and saving the file in an image viewer (recompression). In particular, this means that investigators using standard tools are unable to counteract a deliberate attempt to obfuscate the material.
New Robust and Efficient Methods
To improve search capabilities, Fraunhofer SIT has therefore developed alternatives to existing methods based on robust hashes, which allow content to be recognized even when it has been deliberately manipulated. This means that content can still be reliably identified even when the data has been altered, for example through resizing, targeted content manipulation, format conversions, or mirroring.
Robust hashes from Fraunhofer SIT can be used as a supplement and alternative to previously used hashes when tracking, for example, child pornography, without requiring changes to the rest of the infrastructure. They thus enable greater efficiency in reviewing large amounts of data and allow for the automation of searches for relevant material, making investigative work more efficient.
Fraunhofer SIT offers software for the robust and efficient detection of similar images. The software can be licensed or integrated into other IT forensics products.