image thumbnail

tobycollins/GAIM

version 1.2.0.0 (11 MB) by Toby Collins
Matlab implementation of Graph-based Affine Invariant keypoint Matching (GAIM)

400 Downloads

Updated 9 Apr 2016

From GitHub

View License on GitHub

This is the Matlab implementation of Graph-based Affine Invariant keypoint Matching (GAIM) from the peer reviewed paper "An Analysis of Errors in Graph-based Keypoint Matching and Proposed Solutions" by Toby Collins, Pablo Mesejo and Adrien Bartoli, published in the European Conference on Computer Vision, September 2014. GAIM solves the general keypoint-based graph matching problem for two images. It does not require prior knowledge about the number of objects, the amount of occlusion, the amount of non-rigid motion between the images, the amount of background clutter, nor the objects' topologies (which can change if e.g. an object tears). This is the author's implementation. Please cite the paper if you use any part of this code in your work. Above all, we hope that you will find this code useful.
Graph-based keypoint matching is a very difficult computer vision problem, is not yet solved, and requires ongoing research. The problem involves detecting keypoints (or features) in the two images using e.g. SIFT or SURF, and then matching the keypoints using their descriptors and geometric constraints between neighbouring keypoints. Solutions are important for several applications, including non-rigid object tracking, registration, and non-rigid scene understanding. For it to work, the objects in the image must be well textured (so that keypoints can be detected), and the images should not be badly corrupted by noise, blur or compression artefacts. We must also be looking at the same object in the two images. The object can however undergo deformation and be occluded either by itself or by other objects. Any extensions or improvements to GAIM are most welcome.
The entry-level function is GAIM_matcher. Please call GAIM_example() to run GAIM on three example image pairs, given in ./examples ('cardboard', 'desktop' and 'tear').

Setting dependencies and compilation has been made as easy as possible. We include a function to automatically download and compile the necessary dependencies (setGAIMDependencies()). This gets called automatically when you run GAIM_example(). This has been tested on Linux (Ubuntu 14.04 b4bit) and windows (version 10 64bit).

Cite As

Toby Collins (2022). tobycollins/GAIM (https://github.com/tobycollins/GAIM), GitHub. Retrieved .

MATLAB Release Compatibility
Created with R2014a
Compatible with any release
Platform Compatibility
Windows macOS Linux

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!
To view or report issues in this GitHub add-on, visit the GitHub Repository.
To view or report issues in this GitHub add-on, visit the GitHub Repository.