This directory contains programs and images used in paper:
H. Y. Kim and R. L. Queiroz, "Inverse Halftoning by Decision Tree Learning," IEEE Int. Conf. Image Processing, 2003.

The files are divided in three ZIP files:

program.zip - contains source and executable programs (uses old IMG library)
program_proeikon5.zip - contains source and executable programs (uses new Proeikon5 library)
a-set.zip - set of images A
b-set.zip - set of images B

The easiest way to carry on the tests is:
1) Download 3 zip files, uncompress them in the same directory.
2) Run a-run.bat to run the first set of tests.
3) Run a-dist.bat to compute PSNR between images, written in a-dist.txt.
4) Run b-run.bat to run the second set of tests.
5) Run b-dist.bat to compute PSNR between images, written in b-dist.txt.

Program.zip contains the following files. All programs are written in C++ and they must be compiled using DJGPP compiler and IMG library ( http://www.lps.usp.br/~hae/software ). All programs will run on Windows 98, ME and XP. Probably, the programs also run on Windows 95, 2000 and NT.

Program_new.zip must be compiled using Proeikon library.

bldi.cpp bldi.exe     - Builds decision-tree for inverse halftoning choosing peepholes sequentially.
appi.cpp appi.exe     - Applies sequential decision-tree, performing the inverse halftoning.
E88.MAT               - 8x8 window specification to be used by bldi program.
bldo.cpp bldo.exe     - Builds decision-tree for inverse halftoning using entropy maximization (Occam's razor).
appo.cpp appo.exe     - Applies Occam's decision-tree, performing the inverse halftoning.
E88.GES               - 8x8 window specification to be used by bldo program.
distg.cpp distg.exe   - Computes the difference (MAE or PSNR) between two gray-scale images.

A-set.zip contains the following files:

a-run.bat       - batch file to run the test
a-dist.bat      - batch file to compute PSNR
aay.tga         - sample output image (grayscale)
aax-clus.bmp    - sample input clustered-dot ordered dither
aax-disp.bmp    - sample input dispersed-dot ordered dither
aax-errd.bmp    - sample input error diffusion
aax-hpcd.bmp    - sample input HP laserjet coarse-dot
aqy.tga         - ideal output image (grayscale)
aqx-clus.bmp    - image to-be-inverse-halftoned clustered-dot ordered dither
aqx-disp.bmp    - image to-be-inverse-halftoned dispersed-dot ordered dither
aqx-errd.bmp    - image to-be-inverse-halftoned error diffusion
aqx-hpcd.bmp    - image to-be-inverse-halftoned HP laserjet coarse-dot

B-set.zip contains the following files:

b-run.bat       - batch file to run the test
b-dist.bat      - batch file to compute PSNR
bay.tga         - sample output image
bax-clus.bmp    - sample input clustered-dot ordered dither
bax-disp.bmp    - sample input dispersed-dot ordered dither
bax-errd.bmp    - sample input error diffusion
bqy.tga         - ideal output image
bqx-clus.bmp    - image to-be-inverse-halftoned clustered-dot
bqx-disp.bmp    - image to-be-inverse-halftoned dispersed-dot
bqx-errd.bmp    - image to-be-inverse-halftoned error diffusion