‹header›
‹date/time›
Click to edit Master text styles
Second level
Third level
Fourth level
Fifth level
‹footer›
‹#›
Oriented polygonal surface mesh: smooth linear approximation of the surfaces of the objects.  Oriented means the we stor the normal vectors to the surface.
The title of my thesis is Stereo-based Hand Gesture Tracking and Recognition in Immersive Stereoscopic Displays. As the title indicates, this research focuses on developing new ways of interactions using gesture tracking and recognition based on a pair of stereo cameras. At the same time we study the type of displays with which this interaction will occur. I will start with some of the exploration I have done which lead me to this research. Beginning with immersive displays:
The goal of this thesis is to facilitate and increase the ways of communication. Not only between humans, but also between a human and the body of knowledge . By increasing the bandwidth and the intuitiveness of communicating with the digital media, virtual worlds, computers etc...  we facilitate indeed communication among humans and the body of knowledge therefore optimizing the efficiency of acting on the world and therefore
The means that this thesis will address to facilitate communication with the digital medias is improving interactions with improved immersive visual stereoscopic mediums through natural gestural based on stereo tracking. There are other means of course to interact with computers such as speech and the such, but in this thesis, our focus is to improve the visual displays and most of all, the gist is in creating new ways of interacting using hand gestures
Why do we want to do that because
The illusion of immersion is created by  projecting a stereoscopic computer graphics  into a cube made of screens that completely surround the viewer. Viewer walk in this space  bundeled with head trackers, a wand which allows them to interact with the stereoscopic virtual environment.
The current state of current displays and interactive tools is limited. Indeed a mouse and a keyboard provide limited efficiency when it comes to manipulating 3D objects. Monitors are flat 2D screens with a limited FOV. Joystics and mises have few degrees of freedom, a joystic has in average 3 to 4 degrees of freedom as opposed to our had which has 27 degrees of freedom and is more natural to use to manipulate objects, therefore more intuitiveness would be added by use of our hands. That is why Gloves and styluses as the phantom which provide a higher degree of freedom has been used, but these have the disadvantage of having a physical connection. Trackers as the polymous also have been used, but these are electromagnetic tracker that needs a combersum physical connection. Likewise for head mounted displays who provide a better ...
We need to create a system, that has the advantages of a wide field of view, high degree of freedom, stereoscopic, and yet be intuitive and not requiring any combersom physical connections...
In order to accommodate for the shortcomings of the previous technologies we presented we imagined and built the following system that provides an immersive visual environment that is high resolution not needing any physical Heavy Head Mounted apparel nor combersom wires. The system now is 3 PCs
The rest of this presentation will be describing our work in visual displays and then the main part the stereo-based
Here is our proposal, some of these points has been done, some are to be done, we will go and detail each of these points in the next slides, after we go through all the points, we will at the end advocate why we chose to use stereo matching in tracking and recognizing gestures and compare it to other methods that has been implemented in the litterature.
In order to create such a system, we need to implement and improve the following topics:
Camera Calibration.
Stereo matching and reconstruction of the hand
Hand modeling.
Initial Pose of the hand model.  (using the stereo reconstructed hand).
Tracking of the hand (that is determination of the position and orientation of the hand in real time).
Building a gesture library. Which is intuitive and easy for the user to use.
Gesture recognition and interaction with the 3D world by use of the tracked hand.
The remaining of our presentation will be taking each of these points in our proposal and talk about the competing method in the litterature, then our method describing the differences, advantages and disadvantages and so forth. We start with camera calibration:
The major uniqueness of our approach is that we are using stereo to reconstruct the hand