Web-Based Image Forensics For Catching Child Predators

Sam Khoury (Minneapolis, MN) -

In just a short time, the Internet has created huge economies of scale for information dissemination, affecting everything from how news spreads among bloggers, journalists, and average readers to how consumers compare sale prices among stores for their holiday shopping. The dark side of such information efficiency, however, is that global criminal elements are exploiting it in various ways. From utilizing virtual communities to talk to each other to going unnoticed under the cover of anonymity, criminals of all types are deepening their relationships with each other and becoming more powerful.

The good news is that this exploitation pendulum swings both ways. Here, I discuss the particular case of those bad actors commiting crimes against children, and how law enforcement is leveraging new Web-based digital image technologies combined with cloud computing to combat the evil work of child predators.

Scarcity to Overload: A Brief History of Information

Once upon a time, not so long ago, I had to call a broker on my phone (a “land line”) in order to get information about my investments. Time on the phone was rushed, with a long list of company names and stock prices rapidly spewed at me - after all, my broker had other calls from other clients to answer. Information was scarce, and getting it depended on gatekeepers.

Now, in the Web 2.0 age, there are relatively few information gatekeepers. Today I use my smart phone to pull up screens full of ticker symbols showing real-time information about stocks, bonds, commodities and exchange rates. My brokerage beams them to me like Scotty from Star Trek. Information is “free” to the point where we are practically overloaded with it. We need filters to help us figure out what information is important, or necessary - and what isn’t.

This abundance of information is both a blessing and a curse, an opportunity and a risk. Society’s newly acquired efficiency in disseminating data, information, images, audio, and video has not gone unnoticed or unused. Criminals and other bad actors like terrorists are now interconnected in loosely joined but powerful social networks that afford both identity and anonymity, which complicates matters for those trying to infiltrate and counteract them. In some ways, the Internet and its massive information dissemination capabilities have empowered and emboldened criminals. Those acting against children, whether it be kidnapping, pornography, prostitution, or similar crimes, are a particularly noxious case to consider.

Child Exploitation Meets Information Overload

Internet-based sexual abuse of children takes many forms, including the distribution of sexually abusive images. The global production and distribution of such images is a difficult problem to fight because of the extreme nature of the offense - Not only do offenders go to great length to hide their identity, but the rapid alteration and redistribution of images around the world makes it almost impossible for the relevant authorities to identify the origin of the geographic location where such abuse took place, and therefore harder to identify and help the victims or apprehend and punish the perpetrators.

If we further examine both the explicit and implicit limitations imposed on law enforcement when it comes to dealing with such a crime, we see that the problem is in fact even more complicated. Not only do law enforcement agencies fight Internet crime with outdated tools in many cases, but there are serious jurisdictional issues that play a major role in reducing the efficiency of combatting this type of offense. This is because the use of the Internet to distribute abusive material is not jurisdictionally bound. Images can be distributed around the world in seconds, whereas law enforcement resources are compartmentalized along jurisdictional and legislative lines.

This disadvantage has resulted in a sizeable performance gap between the ability of perpetrators to commit such crimes and the ability of law enforcement agencies to combat it. This disadvantage is further compounded with recent explosive growth of the Internet and the constant transmittance of billions of messages, images, videos and other form of communicative material over the global network. Further breaking the problem down, the two most critical issues that face law enforcement are (1) identifying new abuse images and (2) sharing information/intelligence.

(1) Identifying new abuse images: Consider the scenario of a particular law enforcement agency that seizes one or more computers containing child abuse material. Potentially thousands of images may have been stored on those computers, and every image must be examined, identified, categorized and checked against images that exist in the agency’s database. Aside from the psychological toll inflicted on investigators who repeatedly handle such material, there exists a massive efficiency problem in processing thousands of images. Further, the process of image distribution may involve cropping, resizing, rotating, or otherwise altering images, making the job of law enforcement exponentially more difficult.

(2) Information Sharing: The 9/11 commission stated that the biggest impediment to all-source government intelligence analysis is the human or systematic resistance to sharing information. Due to the sensitivity of both the materials and investigations, along with a multitude of restrictive operational policies, law enforcement agencies tend to oppose sharing information about child exploitation. Unfortunately, this approach also results in duplication of effort amongst multiple agencies and even countries investigating the same case.

However, as the pendulum swings back, emerging technologies are now beginning to enable law enforcers to combat the new wave of high-tech crimes via the Internet.

Digital Image Forensics to the Rescue

Forensic technologies for digital images have now advanced to the point where they are becoming invaluable for combating child exploitation.

In an attempt to bridge the aforementioned technological gap between perpetrators and law enforcement, researchers have been working on enhancing the imaging technology in a variety of ways. One of the most intriguing and promising image forensic techniques is known as “image fingerprinting.” One example of this is software named PhotoDNA, an imaging technology using a technique called “robust hashing” to produce highly unique and reliable digital signatures - In other words, PhotoDNA can produce the equivalent of a human fingerprint for images.

PhotoDNA was recently donated to the National Center for Missing & Exploited Children in order to more efficiently counteract the proliferation of sexual abuse images. Brad Smith, the general counsel for Microsoft, commented on this in a press release:

PhotoDNA is a powerful technology that will help combat online child pornography. We are very pleased to donate PhotoDNA to the National Center for Missing & Exploited Children to support the battle against this horrific crime.

In distinct ways, this same technology will also add to the arsenal of technological tools employed via law enforcement to respond to challnges in the area of global child exploitation. Now that any given image can be associated with a highly unique and reliable signature, identification of new images becomes a sigificantly simpler task.

Implications of Digital Signature Technologies

There are a number of implications of digital signature technologies like PhotoDNA becoming part of the law enforcement process. The first is simple: time savings. When every image is transformed into a digital signature and stored in a shared database as such, all newly obtained signatures can be automatically queried against the database. The introduction of an automated image signature search capability will help abandon any labor-intensive process resulting in quickly identifying existing images under investigation and identifying new victims. In concept, this is identical to the Automated Fingerprint Identification System (IAFIS) used at the FBI.

The second implication is what one might call “the ratchet,” in that the process of obtaining a digital signature is irreversible - once you have it, you have it, and the number of signatures in a database can only increase. In turn, the more images in the database, the more useful it becomes to everyone who uses it. Further, PhotoDNA can even account for very minor changes to the same image, and therefore the “ecosystem” of images around a victim, a criminal, or a case can be better understood as a system of complex information sharing.

Finally, the use of digital signatures implies a different means where information about child sex crimes can be shared. Because, like actual fingerprints or biological DNA signatures, one can have evidence that two people are the same without actually seeing what they look like, sex crimes information can be “shared without being fully shared,” which is respectful of victims, does not violate certain local laws, and has other useful qualities.

In the future, a global law enforcement community could be established in order to host a cloud-based repository for such digital signatures that can be queried with new images from the community.

INTERPOL and Microsoft: What’s Next?

The International Criminal Police Organization, or INTERPOL, already has such a database for actual images. Software like PhotoDNA takes this concept to the next level.

There are challenges, though. One is that the contents of the INTERPOL database must be submitted via member countries. However, one can imagine that many crimes take place in some countries that are not necessarily working with INTERPOL, and thus this shared database model has probably not reached its full potential in fighting crimes related to child exploitation. Nevertheless, in the future such a model may encourage more global organizations to actively contribute to populating the database.

Large technology companies have a role to play as well. Microsoft works extensively with the National Center for Missing & Exploited Children (NCMEC) as well as the International Centre for Missing and Exploited Children (ICMEC) to help in their efforts. Additionally, Microsoft has established a Digital Crimes Unit to advance relevant technologies, enhance enforcement, and establish partnerships that can help make the Internet safer for everyone.

Technology is not a magic bullet. Fighting internet-based sexual abuse of children is a multi-faceted approach that involves complex technological, organizational and legislative issues. Unfortunately, the fight against child exploitation in the age of the Web will be a significant challenge for years to come.

Sam Khoury is a Public Sector Solutions Director for Microsoft.

Image of INTERPOL’s logo from Wikipedia. Images of stock ticker, a child, CSI, and a digital signature all used under Creative Commons.

Twitter Digg Delicious Stumbleupon Technorati Facebook Email

5 Responses to “Web-Based Image Forensics For Catching Child Predators”

  1. scholarships for women Reply Dec 23rd, 2010 at 9:35 pm

    Nice site, nice and easy on the eyes and great content too.

Trackbacks/Pingbacks

  1. Anonymous - Dec 7th, 2010

    [...] This post was mentioned on Twitter by Dr. Mark Drapeau and Regilberto Girão, jack d. jack d said: RT @cheeky_geeky Web-Based Image Forensics For Catching Child Predators - http://is.gd/ilANu (via @SectorPublic) [...]

  2.   Law Custom Research Paper|Asbestos Attorney Washington - Dec 9th, 2010

    [...] Web-Based Image Forensics For Catching Child Predators | SECTOR: PUBLIC [...]

  3. Web-Based Image Forensics For Catching Child Predators | SECTOR … | StopAPredator - Dec 11th, 2010

    [...] [...]

  4. Online Internet Dating Predators www.PersonalSecurity.TV | Online Dating Tips - Dec 18th, 2010

    [...] Web-Based Image Forensics For Catching Child Predators | SECTOR: PUBLIC [...]