VORONOI ORDERED SPACE - Video Object Extraction and Representation: Theory and Applications - page 63

Figure 3.10. Levels of Contour Description: The initial estimate of shape is shown
as the grey blob in the center of images; the true shape is shown as darkened contour.
a) A set of curves, b) the subset of curves that are smooth, c) the subset of curves
that are also equidistant, d) the subset of curves that are smooth, contains the initial
estimate, and are Voronoi Order Consistent. Note that most restrictive set of curves

that still contains the true contour is d).

7. APPLICATIONS
We use Voronoi Ordered Spaces in two applications in this topic:
Video Object Segmentation in support of the MPEG-4 standard and
Image/Video Object Query by Shape in support of MPEG-7 standard.

APPLICATION #1: RESTRICTING OBJECT
BOUNDARY SEARCH SPACE
In Chapter 4, the Voronoi Ordering provides a seamless interface of
a priori shape information for our video object extraction system. As
shown in Figure 3.10, if we have an reliable initial contour estimate,
Voronoi Ordered Space can be used to restrict the search space of an
object boundary search. In the most general and least restrictive case,
we use smoothness. If we enforce that a boundary estimate must com-
pletely contain the initial estimate, we can further restrict our search
space. If we assume our estimate has the correct shape of the object,