View source

From jderobot

You do not have permission to do that, for the following reason:

The action you have requested is limited to users in the group user.

You can view and copy the source of this page:

* '''Project Name:''' 3D Reconstruction by means of visual attention system

* '''Authors:''' Roberto Calvo Palomino (rocapal [at] gsyc [dot] es) and Jose María Cañas Plaza (jmplaza [at] gsyc [dot] es)

* '''Academic Year:''' 2007-2008

* '''Degree:''' Grad

* '''Jde Version:''' [http://svn.jde.gsyc.es/jde/jdec/tags/4.2.1/ jde-4.2.1]

* '''SVN Repository:''' [https://svn.jde.gsyc.es/users/rocapal/attention https://svn.jde.gsyc.es/users/rocapal/attention]

* '''Tags:''' vision, attention, 3D, saliency, 	correlation, convergence

* '''Technology:''' c, c++, jde suite, openGL, opencv

* '''State:''' Finished

* '''Source License:''' [http://www.gnu.org/licenses/gpl-3.0-standalone.html GPLv3]

* '''Document License:''' [http://creativecommons.org/licenses/by-sa/3.0/ Creative Commons Attribution-Share Alike 3.0 Unported License]

* '''Abstract:'''
This current project develops a visual attention system for a mobile robot that generates representation of 3D environment. For this, the attentive control is based on a map of salience on the relevant objects in the image. Through a pair of cameras and analyse of the real images, we can such localize relevant objects in the 3D world. The system attentive scheduled analyzes the image in a way similar to what makes a human eye, looking at things that focus to attention: colors, edges, motion or prior knowledge of the objects in reality. Some theories argue that the human visual system ihas limited capacity of processing, and the attention acts as neuronal filter that selects the information, should be procesed in every moment.

* '''Documentation:'''

 Masther Thesis: [http://jde.gsyc.es/multimedia/pfc/rocapal/grad/Visual_Attention_3D_2008.pdf PDF] - [http://svn.jde.gsyc.es/users/rocapal/attention/trunk/memoria/ Docs_Sources]


* '''Videos:'''


==='''3D Reconstruction'''===


The system works fine. The system represents in 3D a complex figure, as a hand.

<video width=800 height=470 >http://jde.gsyc.es/multimedia/pfc/rocapal/grad/videos/reconstruccion3D_Mano_24-08-2008.flv</video>


The system is autonomous. The system represents the information of the cameras in 3D. In the next video you can see the representation in 3D of the scene.

<video width=800 height=470 >http://jde.gsyc.es/multimedia/pfc/rocapal/grad/videos/reconstruccion3D_8-08-2008.flv</video>

==='''Cognitive Saliency'''===


Added new feateure. A cognitive saliency can make assumptions about the fourth corner of a square.

<video width=800 height=470 >http://jde.gsyc.es/multimedia/pfc/rocapal/grad/videos/cognitiveBehaviour_6-09-2008.flv</video>

==='''Tracking objects'''=== 


In this example the system follows one object.

<video width=800 height=470 >http://jde.gsyc.es/multimedia/pfc/rocapal/grad/videos/followObjectAttention.flv</video>


In this example the system follows two object.

<video width=800 height=470 >http://jde.gsyc.es/multimedia/pfc/rocapal/grad/videos/follow_2_ObjectsAttention.flv</video>


==='''Attention system'''=== 


You can see in the next video how the movement of the cars is followed by the "virtual eye". The system works with saliency map that decide the target. 

<video width=800 height=470 >http://jde.gsyc.es/multimedia/pfc/rocapal/grad/videos/attention_movement-final.flv</video>




* '''Evolution:'''

<img width="640" height="480">http://jde.gsyc.es/multimedia/pfc/rocapal/grad/evolution_rocapal.png</img>


* '''More information:'''

Return to Rocapal visual attention 3D.

Retrieved from "http://jde.gsyc.es/index.php/Rocapal_visual_attention_3D"

View source

From jderobot

Views

Personal tools

main

projects with jderobot

docs

developers

communication

Search

Toolbox