Head-Tracking Pointer
An application that, using an inexpensive camera, lets users control a mouse pointer by aiming their face around the screen.
Date Posted: August 23, 2005
|
|
 |
|
|
Update: June 20, 2006
New version includes several bug fixes and minor changes to the algorithms that make the application work better.
What is the Head-Tracking Pointer?
The Head-Tracking Pointer provides an inexpensive and easily-used mouse replacement for those unable to use traditional pointing devices. Using only software and any Web-cam, this application allows users to point and click with character-level accuracy by simply aiming their face.
Unlike other head-tracking solutions, the Head-Tracking Pointer does not require special-purpose hardware, head gear, or tracking aids, such as retro-reflective dots. Users train the system by tipping their head left and right. Custom-tracking algorithms provide reliable operation in a wide range of environmental conditions. Head motion is translated into pointer motion by custom filters that provide both high speed for long movements and high accuracy for fine control, without switching modes. All clicking variations are supported, including right clicking, double clicking, and click-and-drag.
How does it work? The Head-Tracking Pointer uses state-of-the-art computer vision technology to turn user's head movements into the movement of the mouse pointer.
To use the Head-Tracking Pointer, users connect a camera to their computer and place it near the screen, aiming back at their face. They train the system by tipping their head left and right, then simply aim their face where they want the pointer to go, as if they had a very long nose. Clicking is provided by holding the pointer still for a brief time. The type of click generated is controlled by a menu in the corner of the screen.
The unique motion of the head tipping helps the Head-Tracking Pointer to isolate the user's face from other objects in the camera's field of view. It then selects and memorizes a feature-rich region such as around the eyes or nose. This region will be tracked in subsequent video images so that the system can accurately follow the motion of the user's head. The motion of the head with respect to its original location is combined with information such as the size of the user's face in the image and assumptions about the location of the camera with respect to the screen in order to determine a target on the screen, located approximately where the user's nose is aimed. This location is somewhat coarse due to the limitations of video technology, so it is combined with the pointer motion history to determine the new pointer motion at each moment. The combination algorithm, referred to as the "transfer function," is designed to give both high accuracy during short, fine-positioning movements, and fast response during large, long-distance movements. By adjusting the parameters of the transfer function, users can adjust the pointer movement to suit their needs and abilities.
Further information is available in the IBM Head-Tracking Pointer User's Manual.
|
|
|
| | About the technology author(s):
 Rick Kjeldsen is a research staff member at IBM's T. J. Watson Research Center. He received his Ph.D. from Columbia University in 1997; his M.S. from the University of Massachusetts in 1987; and his BSEE, With Great Distinction, from Clarkson University in 1981. Between stints at school, Dr. Kjeldsen has held various programming and research positions within IBM. He was a member of the Exploratory Computer Vision Group at Watson Research from 1987 through 2001, when he joined the Accessibility Research Group. He has published and filed patents on a wide variety of topics in artificial intelligence, computer vision, and perceptual user interfaces. He is the inventor of the TouchFree Switch, an inexpensive vision-based input device that allows users with limited physical mobility to trigger computer events. Dr. Kjeldsen's image-processing software was featured in the theatrical production Alladeen, which has just completed a world tour. He is the inventor of the user vision-based interaction system for the Everywhere Display, which allows interactive displays to be projected onto any surface. His current research is aimed at the application of human-centered computer vision to create practical, customizable user interfaces for both general and special-needs users.
|
|
|
|
| |
