Welcome!



This site is dedicated to the Shape-Matching Framework project, implemented by Yanir Taflev,
as part of a final academic work in college.

Recent Announcements

  • Version 1.0 is now Available A new version , with final tunings, new Combination Algorithm and many other new featuresis now available in googlecode download page.Enjoy, and don't forget to leave a feedback
    Posted Oct 19, 2009 3:14 PM by Yanir Taflev
  • New public version 0.2 including new Unit Test GUI New public version is now available for download,You can go to project's googlecode download page and download the newest shiny 0.2 version and a full guide.
    Posted Sep 30, 2009 6:41 AM by Yanir Taflev
  • Planet Source Code publications The algorithms are now available in PlanetSourceCode site, here are the links:PCA algorithmHausdorff distance algorithmShape context
    Posted Oct 3, 2009 10:25 AM by Yanir Taflev
  • Code Project Publications The algorithms are now available at CodeProject, here are the links to the articles:PCA articleHausdorff distance articleShape Context article
    Posted Sep 30, 2009 6:29 AM by Yanir Taflev
Showing posts 1 - 4 of 4. View more »



Background:
The Shape-Matching framework designed to provide algorithms so as the infrastructure to make it easy
to align and match shapes.
There are few main algorithms, each one uses different attitude and complement 
other aspect of shape matching issue.
The framework currently in development stages, so feedbacks are more than appriciatable. 

License:
The material introduced in this site is Open Source code.
The code distributed under the name "Shape matching framework" is released for use
under the AFL v3.0 (Academic Free License) terms.
For further information and terms see here.

Innovation:
This project innovates in few subjects over other existing project:
- This code is fully implemented using .NET C# language, the first that not using wrapped C++ code.
- The project implements a uniform access to the algorithms, making it easy to use inside general applications.
- Both the latest source code and the dlls are available to download.
- Some algorithms are implemented first time outside MATLAB as Open Source.
- The core implementation fully compatible with .NET 3.5 framework.
- Uniform methods supported from .NET 2.0 framework and above.

What is inside:
There are three main algorithms that were implemented in this version:
    - Shape context matching.
    - Principal component analysis alignment (a.k.a. PCA).
    - Hausdorff distance - double sided.
There are implementation of some auxiliary algorithms that one may find useful:
    - Thin Plate Splines (a.k.a. TPS).
    - Hungarian algorithm for perfect pairing.
    - Liniar algebra infrastructure including Inverse, Eigen values and vectors implementation.