Active Voodoo Dolls: A Vision Based
Input Device For Non-Rigid Control
For human-computer interaction, a wide variety of input devices
have been invented. One category of these devices are those which
transform intentional human motion into a measurable analog quantity.
Not only must such a device be accurate, but also intuitive in such a
way that the motion of the device corresponds to the motion which the
human is controlling. Basic analog joysticks and mice have two
degrees of freedom, which correspond to x and y axis of the screen.
Spaceorbs have six degrees of freedom, three translational and three
rotational, which are determined by the position and orientation of a
spring mounted rubber ball with internal sensors. More complex and
expensive devices such as haptic feedback pens and polhemus devices
can also provide more than six degrees of freedom. All of these
devices have the limitation that they only sense rigid, or in the best
case, articulated motion.
Although it is theoretically possible to build a device to measure
more degrees of freedom in motion (bending, twisting, etc.), vision
based techniques offer an inexpensive alternative to control without
the wiring and cumbersome devices.
Active Blobs can be used as an input device for non-rigid
control that can be used for animation and video game applications.
The user grasps a soft, squishable object in front of a camera which
can be moved and deformed in order to specify motion. Active Blobs, a
non-rigid tracking technique is used to recover the position, rotation
and deformations of the object. The resulting transformations can be
applied to a texture mapped triangular mesh, thus allowing the user to control it
interactively. Our use of texture mapping hardware in tracking allows
us to make the system responsive enough for interactive animation and
video game character control.
Readers are referred to the
technical report. for a detailed description of the approach.
Figure 1: Animating a bunny character
via direct manipulation of a spongy object. The system observes the
user manipulation of a deformable spongy object via a camera;
representative frames from such a video sequence are shown in (a).
The system then tracks the motion of the object by tracking a
deformable region as shown in (b). The recovered deformation
parameters are then applied to the graphics model for animation as
shown in (c).
Figure 2: Animating a three-dimensional animal cracker
box via direct manipulation of a spongy object. The video sequence
is shown in row (a). The deformable region used for tracking is in row (b).
In row (c) the deformation parameters are applied to two dimensions of a
three-dimensional animal cracker box.
© 1998 Image and Video Computing Group - Boston University
Page Created by: John Isidoro
Last Modified: Mar 13, 1998