Affine Invariant Feature Matching Using a Grassmannian Graph

PROJECT TITLE :

A Grassmannian Graph Approach to Affine Invariant Feature Matching

ABSTRACT:

Feature matching in 2D and 3D for affine invariant 2D and 3D objects is a long-standing challenge in computer vision. There are two stages to our two-stage technique that allows us to restore the correlation between two disorganised, mutually exclusive feature sets (or "points"). An affine invariant Grassmannian representation, in which the features are all mapped into the same subspace, is ideal in the first stage. The Grassmannian coordinate representations differ by an arbitrary orthonormal matrix, it was discovered. These coordinates can then be used to recover correspondences using basic mutual closest neighbour relations after a Laplace-Beltrami operator (LBO) approximation of these coordinates is applied. We use a huge number of experiments on 2D and 3D datasets in our benchmarks. The suggested Grass-Graph approach successfully recovers massive affine transformations, according to the findings of the experiments.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here