Projects

Robot Partners - Collaborative Perceptual Robotic Systems

The Institute for Robotics and Intelligent Systems (IRIS) 

 

Home
People
Robots
Projects
Publications
Contact

Invariant Features from Interest Point Groups

M. Brown and D. Lowe

This project developed a family of invariant features formed from combinations of interest point groups. By choosing groups of 2, 3 or 4 interest points and using them to define a local coordinate frame, features which are invariant under similarities, affinities and homographies can be formed. We have applied this technique to matching images for object recognition, panoramic imaging and fundamental matrix estimation. 
 
Object recognition using Invariant Features from Interest Point Groups