Photo DNA

Fairfax reports:

Internal Affairs has teamed up with to develop software to detect child sex abuse images.

The PhotoDNA software will be used during the forensic analysis of a seized computer, to identify known abuse images.

Internal Affairs minister Amy Adams said it was common for the ministry’s censorship compliance unit to review more than 100,000 images files on seized computers.

“This technology will make the process much faster. It will also allow a greater level of information sharing with our international partners as more systems come online that use this technology.

“New Zealand is one of the first countries in the world to have access to this technology, which gives investigators another valuable tool to help us in the fight against this problem.”

Generally I’m skeptical of technological “solutions” to child abuse images due to the problem with false positives. Filtering software for example sometimes block legitimate sites.

But this initiative is not like this. How PhotoDNA works is it uses a “fingerprint” of the images, which means only known objectionable images are flagged by automated PhotoDNA scans. More precisely, the basis for PhotoDNA is a technology called “robust hashing,” which calculates the particular characteristics of a given digital image — its digital fingerprint or “hash value” — to match it to other copies (and variations) of that same image.

So basically a human views the image once, and classifies it as a child abuse image, and then creates a hash value of it, so it can be easily detected in future. As I understand it, will not use in its filtering but for when the seize a computer under warrant.

If the computer has 100,000 images on it, they will no longer has to review each one. They can run this software on it, and if it says it find 5,000 matches, then they’ll review those 5,000 only. It is possible there will be some other objectionable images in the other 95,000 – but for prosecution purposes they don’t need an exact count.

Some general info on PhotoDNA is here. As I said, while I am usually skeptical, a technological solution which doesn’t create false positives is a very good tool to be using.

Comments (12)

Login to comment or vote