QuickTopic free message boards logo
Skip to Messages

TOPIC:

Segmentation of Dynamic Scenes from the Multibody Fundamental Matrix

4
Elizabeth Talor
07-26-2007
07:52 AM ET (US)
QkJYrm A number of universities have awarded her honorary degrees, and she earned a prestigious job on the staff of Detroit congressman John Conyers. In 1988 Roxanne Brown noted: "Thirty-two years after she attracted international attention for sparking the Montgomery Bus Boycott, Mrs. Parks's ardent devotion to human rights still burns brightly, like a well-tended torch that ignites her spirit and calls her to service whenever she is needed."
3
Kristin BransonPerson was signed in when posted
11-12-2002
01:33 PM ET (US)
I like the idea of this paper. It seems to use math cleverly. I worry that estimating the number of independent motions relies too heavily on there being no image noise. Some questions I had: the product in equation (1) should be taken over i instead of k, right? What is a Veronese map? Just stringing out all the components of two points into a vector? Is R^(Mn x Mx) the same as R^Mn x R^Mn?
2
Josh Wills
11-12-2002
12:56 PM ET (US)
This paper definitely has some very nice ideas. I especially like the automatic recovery of the number of motions.

I do wonder, however, about the applicability of this approach. This work seems to require that ALL correspondences are correct. This is tricky for most scenes but even more difficult for objects that are not roughly planar - the type of objects that are needed to specify their own fundamental matrix. In other works do the author try to detect objects for which a homography would be a better model?
Edited 11-12-2002 12:57 PM
1
Andrew Rabinovich
11-12-2002
05:19 AM ET (US)
Very very interesting. I have a question about feature selection for every trackable object. How are the features chosen for each particular object, and how many are necessary to determine the motion of a given object?

Print | RSS Views: 916 (Unique: 541 ) / Subscribers: 0 | What's this?