This is a MATLAB implementation of the hypergraph matching algorithm for multi-source image correspondences.
This algorithm is published in the paper: Game theoretic hypergraph matching for multi-source image correspondences.
- sift win32
Execute siftWin32.exe
software to install SIFT libraries on Windows system (32-bit/64-bit).
-
Prepare several multi-source image pairs in a folder (e.g.,
images/
). -
Set parameters (
distRatio
,imresizerate1
andimresizerate2
) inrun_matching.m
(see comments). -
Run
run_matching.m
to match key points of multi-source image pairs via hypergraph matching algorithm.
Notes:
-
Tuning
distRatio
to a larger value in [0, 1] will generate more key points and lower the matching speed. -
The reported running time does not include the establishment of hypergraphs and the calculation of edge weights.
Please cite the following paper if you used this implementation in your work.
@Article{zhang17hypergraphmatch,
title = {Game theoretic hypergraph matching for multi-source image correspondences},
author = {Zhang, He and Ren, Peng},
journal = {Pattern Recognition Letters},
volume = {87},
pages = {87--95},
year = {2017},
publisher = {Elsevier}
}
Please report an issue if you have any question about this repository, I will respond ASAP.
Please star this repository if you found its content useful. Thank you very much. ^_^