scrapbook: Canadian develops nose-driven mouse

September 21, 2004

Canadian develops nose-driven mouse

By Lester Haines

There is apparently some good news today for all those readers who enjoy surfing web porn but would prefer to use their right hand for something other than operating a mouse - a Canadian visionary from Ottawa's Institute of Information Technology has developed a nose-operated mouse which looks set to redefine the human-computer interface paradigm.

Dmitry Gorodnichy's "nouse" tracks the tip of the nose to control your cursor and accepts two blinks of the right or left eye in lieu of the standard button click, the New Scientist reports. Gorodnichy hopes that the nouse will make PC use easier for the disabled, and further reckons that it might "also provide more intuitive ways for people to explore computer-generated environments or play three-dimensional video games".

How does it work? Well, it's all about webcams. In 2D mode, the nouse uses a single webcam which initially takes a picture of the user and isolates the 25 pixels which represent the tip of the nose. It then tracks this pattern. Motion detector software checks for blinking, and translates this into clicks.

For 3D orientation, two webcams track the user's snout and cunning software works out the distance from the PC, blah blah blah.

Excuse our sudden loss of interest in the scientific background to this earth-shattering innovation, but it is without a doubt the most singularly pointless waste of technology since someone bolted two wheels to a pogo stick and called it a Segway. While we concede that it may have some future for the disabled computer user - if it actually works, which we sincerely doubt - it will certainly play absolute havoc with hayfever sufferers' Word documents and habitual nose-pickers' email compositions.

We're sure that readers can add a hundred more reasons why the nouse is bound for techno oblivion. Nonetheless, in the interests of balanced journalism we feel obliged to quote Charles Cohen, vice-president of R&D at Cybernet Systems, who asserts: "The 3D nose tracker will definitely have a place in human-computer interaction in future, but most likely in conjunction with the mouse and keyboard rather than as an alternative."

A sound bit of bet-hedging there. As for El Reg, we're more inclined to agree with Joe Laszlo of Jupiter Technology who tells the New Scientist: "I cannot ignore the high silliness factor of nouse. People baulk at doing things that require them to look silly and there is ample room for looking silly here."

The prosecution rests, your honour.

[original article]

Referenced New Scientist article:

Nose-steered mouse could save aching arms

Tired of using a mouse to control your PC? Perhaps there is another option for when your arm starts to ache: your nose. A novel PC control system lets users nudge a cursor around the screen with gentle movements of their nose. Blinking the left or right eye twice takes the place of left or right mouse clicks.

The inventor, Dmitry Gorodnichy of the Institute of Information Technology in Ottawa, Canada, calls his nose-steered mouse a "nouse". In addition to giving people a change from the keyboard and mouse, he hopes it will make using a PC easier for people who have a disability.

And it could also provide more intuitive ways for people to explore computer-generated environments or play three-dimensional video games, he says.

Nose mouse in 3D mode

The nouse takes advantage of webcam technology. For ordinary, two-dimensional applications like a word processor or drawing program, the nouse works with a single webcam, plugged into a PC's USB port.

Tracking software monitors the image from the webcam to work out where a user's nose is pointing, and generates signals that move the cursor round the screen. Meanwhile, motion detection software works out which eye is blinking to simulate a mouse click.

At the start of a session, the nouse's camera takes a snapshot of the user. From this it isolates about 25 pixels representing the tip of the nose and takes readings for the brightness of each pixel. The nouse software then tracks the pattern of pixels.

A double blink switches the nouse on. Its software then scans the region where the nose was last, looking for the 25-pixel target. Once it has found the target, its movements are translated into the same signals that would be expected from a normal mouse.

Gorodnichy has designed several software packages to demonstrate the nouse, including NousePaint, which allows you to draw patterns on the screen using only the nose and eyes (see video).

The nouse can also be used to navigate around 3D computer software, such as virtual design environments and games, but this requires two webcams. In 3D mode, both cameras pinpoint the tip of the nose and the nouse software calculates how far away the user is, and whether they are moving into, or out of, the environment.

Previous face-tracking user interfaces have used the mouth or eyebrows as tracking points. But because these facial features look entirely different when tilted only slightly to one side, software trying to track them can easily become confused.

The tip of the nose is easier to track, Gorodnichy says, because it is possible to see the characteristic pixel pattern even when the nose is rotated. "There is something special about the nose," he says.

Others agree. "The 3D nose tracker will definitely have a place in human-computer interaction in future, but most likely in conjunction with the mouse and keyboard rather than as an alternative," says Charles Cohen, vice-president of R&D at Cybernet System in Ann Arbor, Michigan, which makes gesture recognition systems for TV weather forecasters. He says future computer games might rely on joystick input, but might also allow extra moves driven by the nose.

But Joe Laszlo, a technology analyst at Jupiter Research in New York City, is sceptical. "I cannot ignore the high silliness factor of nouse," he says. "People baulk at doing things that require them to look silly and there is ample room for looking silly here."

Journal reference: Image and Vision Computing (vol 22, p 931)