Cutting Edge: Futures - Is no nudes good news?

If there's one subject which gets people hot under the collar, it's web censorship. Most of us know there is material out there we wouldn't want our children to see. And most companies would not want their employees spending the day browsing the sweaty-palm variety.

Systems like NetNanny and CYBERSitter, which automatically bar access to designated web resources, are a partial answer, but now there is software which attacks the problem in a different way, automatically analysing images for pornographic content.

Pictures must be boring if you're a computer. Software must scan the image pixel by pixel laboriously searching for patterns among the colours.

Imagine if I were to describe a complete image in terms of rows of numbers representing pixel values of the image's colour palette. How could you determine what the image was? It's a problem, which has taxed image-processing researchers for decades. But there have recently been developments.

Microtrope recently released ImageCensor, a program it claims can scan an image displayed on a PC and determine whether it is pornographic. If it is, the program can store a thumbnail of the screen, together with the time and date, sound an alarm and lock the computer until released by a password.

Having evaluated ImageCensor by viewing images from the web, I'm impressed; it does a good job of spotting pornographic images. Microtrope's Philip Harris is reluctant to describe the algorithms used in ImageCensor but confirmed my suspicion that the program works by grabbing all the screen pixels every so often and analysing the overall colour balance of the image, looking for colour tones that match those of the skin. But there must be more to it, because it rarely decided that an innocent head-and-shoulders was pornographic. Yet when faced with the real thing, it got it right almost every time.

Meanwhile, Margaret Fleck and David Forsyth, at the Universities of Iowa and California respectively, have developed "The Naked People Skin Finder".

Their approach is more complex than Microtrope's and uses a two-stage system. First, a series of filters analyses the colour balance of an image, trying to identify areas with the appropriate tones and shininess for skin; then they attempt matches between groups of such areas against known human limb articulations, stored in a pre-computed database. You can find the details at www.cs.

uiowa.edu/~mfleck.

While the fuss about web pornography has brought research into automatic picture recognition into the limelight, people have been working for some time on the general problem of querying databases of images.

One established system is IBM's QBIC ibm.com> developed in the early 1990s. QBIC enables a user to set up a query by sketching where edges should appear in the image, and what the colour composition should be, as percentages of red, green and blue.

QBIC can then search a database for the best matches.

A more recent approach is multi-resolution image querying, based on wavelet analysis for image encoding and compression (see Futures, PCW May '97).

Numinous Technologies, in Seattle, is developing such a system. At www.

numinous.com you can find an interactive Java applet demo which lets you do a freehand sketch, then searches for a match in an image database.

The results are fast and accurate. Here, I see the seeds of a web search interface of the future: freehand sketches combined with natural language queries. It is an exciting prospect.

However, and I assure you that I am not making this up for journalistic effect, as I was writing this article on my PC, with ImageCensor running in the background, my sound card suddenly hooted: ImageCensor had singled out the Numinous pages as pornographic! OK, so they do contain shades of brown which might resemble a deep suntan, but it does seem rather ironic.

Could Numinous sue for loss of business? Or for defamation?

These are serious questions. As the marriage of PC and TV moves inexorably closer, the web will become as mainstream as the daily newspaper. We shall have to decide how much trust we wish to place in programs which claim to discern, on our behalf, the "meaning" of information.

We might already be standing at the top of a slippery slope, down which we slide at our peril.