A Functional Representation for Graph Matching

The implementation of our work FRGM ([project page] [TPAMI]).

Introduction

This work presents a functional representation for graph matching (FRGM). From the functional representation perspective, the matching between graphs can be reformulated as a linear functional between the function spaces of graphs for general graph matching. Moreover, the linear functional representation map can be viewed as a new parameterization for Euclidean graph matching, which allows us to estimate the geometric parameters and correspondence matrix simultaneously.

Usage

The structure is organized as follows:

your_dir/
  -3rd_party
  -data
  -FRGM-D
  -FRGM-E
  -FRGM-G
  -GM_methods
  -PR_methods

The 3rd_party consists of some dependent codes (Shape Context, geodesic, linear assignment, etc) and can be downlowed here. The data can be downloaded here.

The GM_methods and PR_methods consists of the implementations of the compared methods on general graph matching and Euclidean graph matching with geometric deformations, respectively.

Sometimes the server storing the resources above may be unaccessible, you can also download them from Google Drive or our new work ZAC.

Citation

If you find our work useful in your research, please consider citing:

@ARTICLE{FRGM, 
author={Fudong Wang and Nan Xue and Yipeng Zhang and Gui-Song Xia and Marcello Pelillo}, 
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, 
title={A Functional Representation for Graph Matching}, 
year={2019}, 
volume={}, 
number={}, 
pages={1-1}, 
doi={10.1109/TPAMI.2019.2919308}, 
ISSN={0162-8828}, 
month={},}