* Download the tarball from: http://www.anirudh.net/courses/cse585/project2/src/project2_code.tar.gz and untar it using the following command on any UNIX system: gunzip -c project2_code.tar.gz | tar -xvf - A directory named "project2/" will be created. * To compile, enter the directory "project2/" and type "make" on any UNIX machine with gcc. For solaris, type "make -f Makefile.CSE" (this program was developed and tested on Linux using gcc, and Solaris using Sun CC). * To run, type "./project2 ". Simply type "./project2" to view the available options. The program can take the input image in any valid image format supported by the ImageMagick libraries (e.g., GIF, JPG, PGM, PNG, TIF, etc). For the program to work for formats other than PBM/PGM/PPM, the Imagemagick package must be installed on the system and "mogrify" should be in the PATH. If this is not the case, the program can be trivially modified to use the "GrayImage::ReadFromFile()" function instead of "GrayImage::ReadFromFile2()" that is used by default. e.g. ./project2 images/test.gif -s 2 -n 10 -thr 0.002 -m 5 OR simply: "./project2 images/test.gif" to use the default values. * The program creates two data files "*_I.mat" and "*_R.mat" apart from various image outputs (this is essentially the F_r vector mentioned in the theory). These files can be used to generate the final segmentation by using a separate program if necessary. Another program called "segment" (generated from main_segment.cc) is provided which can be compiled using the command "make -f Makefile.Segment" ("make -f Makefile.SegmentCSE" for Solaris). This program takes the original image and the "*.mat" files generated by the main program as the input and performs region closing on it. It also generates a .tec file which can be viewed using TECPLOT to check the vectors generated (a TECPLOT layout file "vector.lay" is also provided for this purpose). e.g. ./segment images/test.gif test_R.mat test_I.mat -t 0.001 -m 10` * All the images generated by the above program are in either PPM or GIF format and can be easily viewed using "xv" or any other image viewer. -Anirudh Modi- http://www.anirudh.net 4/27/2000-Sat