JOURNAL OF MULTIMEDIA (JMM)
ISSN : 1796-2048
Volume : 1    Issue : 5    Date : August 2006

Unraveling Natural Interfaces for Co-Located Groupware: Lessons Learned in an Experiment
with Music
João Carreira and Paulo Peixoto
Page(s): 18-24
Full Text:
PDF (519 KB)


Abstract
The computer is an ubiquitous element of modern society, nonetheless, human computer
interaction is still rather inflexible. Particularly in local collaborative environments, like office
meetings, the property that the mouse and keyboard exhibit of being a gateway for the individual to
act upon a workspace, has barred the way to the production of co-located collaboration
technologies, because the users have to time-share their actions upon the workspace. Despite
recent developments in touch sensitive multi-user tabletops, we still believe the portability, low cost
and potentially large input area of vision sensors presents the most promising approach to unravel
natural human computer interfaces. We present in this paper our design of a computationally
inexpensive vision based interface that allows multiple users to interact simultaneously with a
single computer by performing hand gestures, which are filmed by a static video camera. This
interface attempts to continuously recognize predefined postures and movements using a
view-dependent method. We also present A.C.O, a co-located groupware application that receives
input from the vision-based interface and allows users around a table to collaborate playing
synthesized music instruments by moving their hands. This prototype gave us important hints on
the more immediate obstacles the technology must overcome.

Index Terms
co-located groupware, human computer interface, gesture recognition, computer vision