23

Swedish National Crime Intelligence

Much has been made of the criminal possibilities of the Internet. There are constant reports on computer hackers, financial fraudsters in the media and so on. The repercussions of the growth of the Internet on serious crime, like child pornography, are horrific to contemplate. This chapter highlights how European police forces’ child protection units are working together and using technology, in the form of knowledge retrieval software, to their own advantage.

Pornography is not an issue that most people feel comfortable with, in spite of the recent Hollywood penchant for films glamorizing certain aspects of the industry. Nevertheless, pornography has been around for many years. It is a massive global industry, the size of which cannot be estimated, but it is safe to say it is a market increasing in size and accessibility, aided significantly by the sudden rise in the use of the Internet for dissemination of material. Pornography magnates no longer need to rely on hard copies changing hands, running the risks associated with customs controls and storing vast quantities of film and photographic images, but have embraced technology, using the electronic superhighway both to peddle and access their wares.

A problem spawned by new technologies

Nowadays, it is a simple process to surf the World Wide Web and find untold quantities of pornographic material using apparently innocent keywords, and then to download stills or video clips on to a computer's hard disk or removable storage system, like a CD or a DVD. Up until now, Internet-friendly pornographers, either suppliers or receivers of child pornography, have had relatively easy lives: Internet jurisdiction has not been entirely resolved across borders, and usage is facilitated by the ease of setting up and accessing web sites from anywhere in the world. However, things have taken an entirely different turn, with a new pan-European project led by the Swedish National Crime Intelligence Unit Child Protection Division, set up using funds from the European Union (EU) ‘STOP’ programme (an EU project with the aim of economically supporting the fight against child abuse) and utilizing technology, this time, to the authorities’ advantage.

It was first recognized that identifying perpetrators and victims from pornographic films and photographs can be difficult and time-consuming following a major seizure of video film in Sweden. The task was exacerbated given the poor quality of the film seized and lack of background information available to the local officers. The Swedish investigating officer met with German police and discovered that the German officers were already aware of most of the material that had been seized in Sweden. This case highlighted the need for an international approach to child pornography, based on a central repository of images.

Swedish National Crime Intelligence Unit – combating crime through management knowledge

In November 1997, the Swedish National Crime Intelligence Division was given the job of establishing and maintaining a central image bank of known child pornography pictures, captured at raids, downloaded from the Internet or otherwise. The objective of the project was to learn more about the origin and distribution of pornographic material, to eventually be able to clamp down on the producers of the images and to stop the abuse of children in this context.

Detective Inspector Anders Persson joined the Swedish child protection unit at the start of the project with the job of supervising the operational value and technical development of an image library and mapping the distribution of the material. The library is able to handle both photographs and video film, which are translated into digitized images for ease of storage and search. Initially, the image bank contained 5600 pictures, all from Sweden.

The knowledge problem

Persson found that the biggest problem once the library was set up was that he still could not compare images found with those found by other forces. There was no search facility for the different images and therefore there was still a requirement for police officers who knew from experience where the different images were stored and all the background information surrounding each picture. The problems were compounded when new officers joined the team, as it was impossible for them to find images matching seized videos or stills.

I tried looking at various search solutions, such as text searching of the images. Text searching is a very good way of finding the right pictures, but the problem is that you must keep a large number of staff just describing the different images. Bear in mind that one CDRom can contain up to 12 000 images and in most cases we are handling hard disks with the capacity of up to 10 Gb and sometimes more. If we had to describe each picture in detail before they were operationally available for searching, it would take far too much time and opportunities would be lost. I decided that the future of a search tool for images could not be based upon any textual pre-work of the images.

The technological solution

During his search for the ideal tool for the recognition of images, Persson came across Excalibur Technologies’ Visual RetrievalWare^®. RetrievalWare^® is a software developer's kit for the creation of powerful image-content matching applications for digital libraries, document imaging, positive identification and so on. Visual RetrievalWare can be employed in a variety of applications with repositories of visual data including corporate, online and government information systems.

Following a demonstration of the software, Persson applied for funding from the EU STOP project and was granted financial support to set up the project. Police forces from Germany, the UK, The Netherlands and Belgium, together with representatives from Europol and Interpol, were invited to a meeting at Europol headquarters in The Hague to discuss the best way to proceed. In May 1998, an invitation was sent out to all EU countries, the Baltic states, Switzerland, Sri Lanka, Thailand and the Philippines to join the project, with the USA, Canada, Australia and New Zealand as observers. Positive answers were received from twelve countries and the project started in June 1998.

‘The first step was to collect seized material from the participating countries, mainly to see if there were any differences in the material between countries, but also to see what Excalibur software could handle.’ Running on a basic Pentium PC, Visual RetrievalWare is able to scan in pictures from CD-ROM, then search its entire database. The software is able to recognize the degree of similarity between a searched and a programmed variable with different degrees of information, based on their colour, shape and texture, and allows users to ask simple questions such as, ‘Have you seen anything that resembles this?’ and ‘Where did it originate?’

Visual RetrievalWare employs Excalibur's adaptive pattern recognition processing technology (APRP) which allows the rapid automatic indexing of digital data from a variety of sources, including paper, electronic and sound, and which is based on the conversion of information to binary patterns. In retrieval, the software searches for an underlying pattern from the digital data, not for specific words or images, which allows for typing errors and enables searching on poor quality images.

Once a search is performed, the images found are ranked and shown in order of similarity compared to existing images using a percentage scale, with 100 per cent showing an exact copy. The terms of reference for searching do not include the identification of the victims involved, but focus on the background information available, e.g. similar colours and textures of walls, carpets and bedding, style, colour and shape of bedstead, shape and number of windows, similar level of lighting, etc.

Police forces taking part in the project can, when they come into possession of new image material, have this compared to already known material stored in the central image bank. This search can give clues as to whether the new images are part of a known series, together with data regarding the origin of the other images in the series etc.

Over the six-month period of the project the library has grown and now stores in excess of 100 000 images from England, Denmark, Finland, The Netherlands, Norway, Portugal, Switzerland and Germany. Results so far have been encouraging, with the unit recommending a single international database to be run by a central agency employing Visual RetrievalWare and proposing that all participating countries set up their own national databases with the same concept. The project has proved that an international reference library for child pornography will result in enormous gains in efficiency for the national and international police authorities. Persson commented that using such technology was really the only answer for the police force to be able to search its image bank and has been a major factor in helping them cope with all the enquiries they have received since the start of the project.

Outcomes

While the technology has not led to hard results in the form of arrests, it has contributed significantly to operations, including the recent Operation Cathedral, which culminated in eighty arrests being made simultaneously across sixteen countries. Persson concludes:

How many arrests or cases have been solved because of our database is very difficult to answer. The benefit is that we nowadays have a far better knowledge of all the child porn images that are available over the Internet. As well as positively identifying suspects from related stored images, we can check easily whether images are being produced by known perpetrators or not. We can very quickly answer a query if an image is old and known or a newly produced one, saving valuable time and manpower and resulting in much more efficient investigations and quicker arrests.

Future challenges

While an enormous step forward has been taken with this project, it is now vital that time and resources are invested by the authorities in keeping up to date, not only with the increasing volume of pornography being disseminated around the world, but also with the differing technologies that are being employed by pornographers. It is also increasingly important that political boundaries are lifted and legislation co-ordinated to best effect across international boundaries in the fight against an increasingly global pornography industry. It is only by working as an international team, as demonstrated in this project, that the police and other authorities can hope to crack down and limit the growth and appeal of pornography.