Nov. 9 2006 2:00PM

Nov. 9 2006 3:00PM

This talk briefly presents two recent works in range image
processing done in my research group. The first one is more
traditional and devoted to fitting quadratic surfaces in range data.
Many geometric fitting problems encountered in computer vision are
of non-linear nature and have a large number of parameters.
In using iterative techniques such as the Levenberg-Marquardt
algorithm we are typically faced with problems of good initial
estimates and high risk of getting in local optimums. We
propose a variable decomposition approach to alleviating these
problems by reducing the number of optimization parameters (without
changing the initial optimization problem). This technique has
been successfully applied to fitting quadratic surfaces in range
data and demonstrated substantially improved performance compared
to early approaches.

The second work deals with synthesis of 3D videos using range
image stream. The fundamental assumption of this approach is
the full availability of range images at video rate. We alleviate
this hard demand and assume that only limited resources of range
images are available, i.e. range images exist for some, but not all,
color images of the monoscopic video stream. We propose
to synthesize the missing range images between two consecutive
range images. Experiments on real videos have demonstrated very
encouraging results. Especially, 3D videos were generated from 2D
videos without any sensory 3D data available at all. In a quality
evaluation using an autostereoscopic 3D display the test viewers
have attested similar 3D video quality for our synthesis
technique and rendering based on range data ground truth.