toon-members
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Toon-members] bug in ransac_estimators.h ?


From: Ethan Eade
Subject: Re: [Toon-members] bug in ransac_estimators.h ?
Date: Thu, 29 Jan 2009 00:15:10 -0800
User-agent: Thunderbird 2.0.0.19 (X11/20090105)


Yes, it was an old implementation of mine of the algorithm from 'Invariant Fitting of Two View Geometry or “In Defiance of the eight point algorithm”' (Torr, Fitzgibbon 2004). I believe it worked then, but I make no guarantees about the current version.

In any case, you're probably better off using the five-point algorithm (Nister 2003), which will work with planar scenes. I never got around to adding that code to libtag, though.

- Ethan


Gerhard Reitmayr wrote:
Hi,

Thanks for the heads-up, yes it should use the functions and I fixed it now.

be warned that the essential matrix estimator is not well tested. it was integrated from other code and some reports say it doesn't work :/

best regards,
Gerhard

On 29 Jan 2009, at 07:06, <address@hidden> wrote:

Hello,

Thanks for the great libraries. I may have caught a small bug, unless I am
not understanding correctly. In tag/ransac_estimators.h, line 26:

TooN::Vector<2> dm = m.b-e;

Should this be:

TooN::Vector<2> dm = second_point(m)-e;

? (And similarly elsewhere in that function.)

Best,
Jonathan Ventura
Department of Computer Science
University of California
Santa Barbara, CA 93106-5110



_______________________________________________
Toon-members mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/toon-members

--
Gerhard Reitmayr
MIL, Engineering Department, Cambridge University
http://mi.eng.cam.ac.uk/~gr281/




_______________________________________________
Toon-members mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/toon-members





reply via email to

[Prev in Thread] Current Thread [Next in Thread]