readme

A salient-boundary extraction software package based on the paper: S. Wang, T. Kubota, J. M. Siskind

    RRC (VERSION 2.0): A SALIENT BOUNDARY DETECTION TOOL BY
           COMBINING BOUNDARY AND REGION INFORMATION

This software package is developed for detecting salient closed
boundaries from an image. The main codes were implemented
in C++ using gcc-4.1 in Linux. It also needs the recent Matlab
Software (with Image Processing Toolbox) to perform some image 
preprocessing and result display. This work is covered by Copyright 
(c) University of South Carolina. It incorporates Blossom4 code
for graph matching that is developed by W. Cook and A. Rohe. The 
technical details of this package can be found in the enclosed 
file "document.pdf".

Authors: Joachim Stahl and Song Wang

Update: 08/11/2006


                              INSTALL

After unpacking the package, in the RRC folder, run
     ./install
This will create an executable RRC.


                              USAGE

   (1) Go to the folder "matlab", in the Matlab environment, run
           preprocess('path/foo')
to preprocess the image "foo.pgm" and write the output to "path/foo.line
s". To preprocess all .pgm files in 'path', use the function 
batchPreprocess.m instead,
           batchPreprocess('path')

   (2) From this installation folder, from the shell, run  
          ./RRC path/foo
to extract one salient boundary. We also provide the following choices:

-l #    -   Set Lambda (default 0.0), a real positive number that weights
            the continuity (curvature) term.
-a #    -   Set Alpha (default 1.0), the exponent for the gap lengths.It 
            can take any real value. Recommended range between 1.0 and 2.0. 
-n #    -   Set the number of produced boundaries (default 1).
-h      -   To use intensity homogeneity instead of region area. 

   (3) Go to the folder "matlab", in the Matlab environment, run
            displayResults('path/foo', n)
to display the n detected boundaries (numbered from 0 to n-1), which are 
also saved as EPS and PNG images in the folder "path/". 

A set of test images used in the enclosed "document.pdf" are included in
the folder "images".

                        CONTACT & BUG REPORT
Song Wang
Computer Vision Lab
Department of Computer Science and Engineering
University of South Carolina
Columbia, SC 29208
Email: [e-mail]
TEL: [phone]
FAX: [phone]
http://www.cse.sc.edu/~songwang