I See What You're Signing
Gesture interfaces are putting the hearing in touch with the deaf
From: IEEE Spectrum - 08/10/2005
By: Willie D. Jones

A gesture interface is being developed by Georgia Institute of Technology
researchers as a tool for communication between the hearing and the
hearing-impaired. The Georgia Tech team is collaborating with cognitive
scientists at the University of Rochester and engineers at George Washington
University to create TeleSign, a one-way American Sign Language-to-English
phrasebook that combines a video camera, wrist-mounted accelerometers, and
machine learning. The camera lens studies the area at the front of the
signer's chest where hand gestures are usually made. The user turns the
system on by clicking a button on the wristband, and turns it off after
signing by clicking the button again; TeleSign then searches its database for
the closest probable English-language matches using hidden Markov models. The
two or three most likely matching phrases are displayed on a portable device,
and the user may either choose to re-sign or select a phrase from the list.
Georgia Tech researcher Thad Starner says TeleSign "limits the vocabulary to
a few phrases - currently about 20 - that are sufficient for a variety of
situations and are most likely to elicit responses like a nod or a point in a
particular direction." Starner's team attempted to streamline the system by
working with GWU's Jose Hernandez-Rebollar on a glove interface designed to
identify signs by recognizing beginning and end hand positions as well as the
movements in between. 

Read the entire article at:
http://www.spectrum.ieee.org/WEBONLY/wonews/aug05/0805nisee.html

Links:
Thad Starner
http://www.cc.gatech.edu/fac/Thad.Starner/

Machine Vision Recognition of American Sign Language Using Hidden Markov Models
http://web.media.mit.edu/~testarne/asl/index.html

The Accele Glove - Instrument and method for translating hand gestures into
speech and text
http://www.gwu.edu/~research/gwnt/accele.htm

Hands That Speak
http://www.cio.com/archive/060105/tl_inventions.html

Jose Hernandez-Rebollar
http://idrt.com/Homepage.php?UID=82&PHPSESSID=03b31b9b1533b3ebdc475882d42e5d4b

