[Software] Augmented Reality
New method of point tracking in a video stream

This innovation is a new method of point tracking in a video stream for 2D / 3D model display.

This new method makes it possible to track one or more set (s) of pseudo-random points detected in an image or in a video stream in real time, and to calculate the position and the orientation of the camera which captured the view of this / these set (s) of points. From this camera position, it is possible to display 2D / 3D models for Augmented Reality, especially in environments with little or no texture information where conventional approaches are of little or no use.

This innovation uses the mathematical methods "k nearest neighbors" and "Geometric hashing".
 

Potential applications

• Augmented reality
• Tracking objects with little or no texture information
• Tracking non-identical but similar patterns (e.g. maps)
• 3D model tracking

 

Advantages of the technology

• Tracking objects with little or no texture information (no need for high quality images)
• Not dependant on lighting conditions
• Detection noise-resistant
• 3D model tracking
• Real-time tracking

 

Status

Software registered with the APP
 

Current state of development

Software developed in C++
Operational demonstrator
 

Laboratory

Urban Architecture Nantes Research Centre

Liming YANG
Guillaume MOREAU
Jean-Marie NORMAND