


	An Efficient Direct Method for Improving visual SLAM
		  G. Silveira, E. Malis, P. Rives


-----------------------------------------------------------------------
			   README.TXT
-----------------------------------------------------------------------


A. DESCRIPTION:
This extended object is composed by two visual SLAM (Simultaneous
Localization and Mapping) results, which are discussed in the paper:

 - in the first demo, the camera is initially on the top of and
   pointing downwards a synthetic Pyramid. Texture images were mapped
   onto the scene in order to simulate realistic situations as close
   as possible. Strong changes in illumination were imposed as well;

 - the second video corresponds to a real-world sequence, which was
   acquired in Versailles, France. Observe the large inter-frame
   displacements.

In both videos, the left frame shows the input images superposed with
the tracked regions, while the right frame shows both the 3D pose and
scene structure being incrementally recovered.

It is emphasized that:

 - bundle adjustment is not performed in any case. The system is
   devised so that real-time constraints can be respected;
 
 - no off-line training phase is carried out. Visual SLAM starts
   immediately after first image arrives;

 - a priori knowledge is not imposed anywhere. For example, that the
   camera moves on a plane, or at a given tilt angle, etc. In this
   aspect, the proposed approach is generic.


B. PLAYER INFORMATION:
The demos were encoded with MSMPEG4V2 codec (Microsoft MPEG-4
v2). They were tested both under Linux with MPlayer, as well as under
Windows with Microsoft Windows Media Player version 10.


C. CONTACT INFORMATION:
Geraldo Silveira (1,2), Ezio Malis (1), Patrick Rives (1)

(1) INRIA Sophia Antipolis - Projet ICARE 
    2004 Route des Lucioles - BP
    93 06902 Sophia-Antipolis Cedex, France
    Phone  : 0033 (0)4 92 38 77
    69 Fax : 0033 (0)4 92 38 50 28
    http://www-sop.inria.fr/icare/personnel/silveira/index.html
    http://www-sop.inria.fr/icare/personnel/malis/index.html
    FirstName.LastName@sophia.inria.fr

(2) CenPRA Research Center -- DRVC Division
    Rod. Dom Pedro I, km 143,6, Amarais
    CEP 13069-901, Campinas/SP, Brazil
    Geraldo.Silveira@cenpra.gov.br


Attachment filename: 1531_VI.mpg (2941204 Bytes)