I. SETUP

Please, read this carefully:

Before running anything you MUST edit the 'smt_setup.m' file and change the variable
'working_dir' to the path to this package (e.g: /<your directory>/Fast-SMT-Design-v0)

All the matlab scripts you may write in the future using the SMT code should 
first execute the ./smt_setup.m script. 


II. RUNNING THE DEMO

We provide a demo script in "smt_faces.m" that performs a similar experiment
to the one presented in the paper "Fast Signal Analysis and Decomposition on Graphs using the Sparse Matrix Transform" (ICASSP 2010).

III. CODE ORGANIZATION
The system is made by code written in matlab (mostly), 
with the core neighborhood smt written in java.

All the core routines are under ./smt/matlab and ./smt/java. 
The Java classes provided are already compiled and placed 
under ./smt/java/FastSMT.jar. 

You should be able to use the SMT in your matlab scripts without having 
to modify or deal directly with the Java routines. The appropriate 
interface between matlab and the Java classes is done inside the SMT 
core routines (under ./smt/matlab) and it should be completely transparent 
to you.

If you desire to modify and recompile the Java portion of the 
code, please, refer to the ./smt/java/build.sh script for further 
details. The code is currently being compiled using Sun's JDK 1.6.0_17,
available for download at http://www.sun.com


IV. FURTHER DOCUMMENTATION

We refer to the following papers to a further description of the SMT:

"Fast Signal Analysis and Decomposition on Graphs using the Sparse Matrix Transform" (ICASSP 2010)


IV. LICENSE AND CITATION

You can use this software free of charge, please read the file LICENSE in this directory 
for further explanation. 

Any puplication that makes use of this code should cite the paper:

"Fast Signal Analysis and Decomposition on Graphs using the Sparse Matrix Transform", ICASSP 2010.


