SpecMatch: A software package for Spectral Graph Matching

Examples of matching obtained using the SpecMatch package.

SpecMatch is an open-source software (OSS) package that performs graph matching using Laplacian embedding and point registration using the expectation miximization algorithm.

SpecMatch falls under GPL, i.e., the general public license. Hence, use of SpecMatch requires a license agreement for all users.

You can freely download the source files (written in C/C++) from this page ( download ).

The graph matching algorithm used by SpecMatch is described in the following paper published at CVPR'08:

Articulated Shape Matching Using Laplacian Eigenfunctions and Unsupervised Point Registration

Publications using any part of SpecMatch should cite the following reference:

@InProceedings{MHKCB08,
  author       = "Mateus, Diana and Horaud, Radu P. and Knossow, David and Cuzzolin, Fabio and Boyer, Edmond",
  title        = "Articulated Shape Matching Using Laplacian Eigenfunctions and Unsupervised Point Registration",
  booktitle    = "Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition",
  year         = "2008",
  url          = "http://perception.inrialpes.fr/Publications/2008/MHKCB08"
}

The interested user/reader may also be interested in the following related publications:

Coherent Laplacian 3-D Protrusion Segmentation

@InProceedings{CMKBH08,
  author       = "Cuzzolin, Fabio and Mateus, Diana and Knossow, David and Boyer, Edmond and Horaud, Radu P.",
  title        = "Coherent Laplacian 3-D Protrusion Segmentation",
  booktitle    = "Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition",
  year         = "2008",
  url          = "http://perception.inrialpes.fr/Publications/2008/CMKBH08"
}

Inexact Matching of Large and Sparse Graphs Using Laplacian Eigenvectors

@InProceedings{KSMH09,
  author       = "Knossow, David and Sharma, Avinash and Mateus, Diana and Horaud, Radu P.",
  title        = "Inexact Matching of Large and Sparse Graphs Using Laplacian Eigenvectors",
  booktitle    = "Proceedings 7th Workshop on Graph-based Representations in Pattern Recognition",
  month        = "May",
  year         = "2009",
  publisher    = "Springer",
  address      = "Venice, Italy",
  url          = "http://perception.inrialpes.fr/Publications/2009/KSMH09"
}

Inquiries should be addressed to mateus AT cs DOT tum DOT edu

Future announcements such as software releases will be communicated via this web site.

---

Last modified: Tue Apr 14 14:33:23 CEST 2009