Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!pipex!lyra.csx.cam.ac.uk!warwick!uknet!festival!leeds.ac.uk!news
From: venkat@scs.leeds.ac.uk (N B Venkateswarlu)
Subject: Annonnucement of testimages for Image segmentation
Message-ID: <1994Dec18.093933.16544@leeds.ac.uk>
Sender: news@leeds.ac.uk
NNTP-Posting-Host: csgi45.leeds.ac.uk
Organization: The University of Leeds, School of Computer Studies
Date: Sun, 18 Dec 1994 09:39:33 +0000 (GMT)
Lines: 91

Information About The Test Images for Segmentation
           ---------------------------------------------------


 In this data base, we include both scanned and artificially generated
 document  images which can be used to test any segmentation  method. Some 
 of the images are used in some articles. We would like to thank Dr.Kamel 
 of University of Waterllo, Canada for providing his images (kamel1.pgm.gz 
 and kamel2.pgm.gz) and to   Dr. Eikwil, Norweign Computing Center, Oslo, 
 Norway,   for   providing    images   map.pgm.gz,  calender.pgm.gz  and
 blue_print.pgm.gz. Other imagesare belongs to our school.  

       All the images are stored in PBM format which can be read using John 
 Bradleys (bradley@cis.upenn.edu) "xv2" package (ver. 2.21). We read all the
 images with this xv2 before packing them in this tar archive.  


     Images test75.pgm.gz, test150.pgm.gz and test300.pgm.gz are of the same
 document with scanning resolutions 75, 150, and 300 dpi respectively. These
 can be used to test sensitivity of segmentation algorithms for varying scan-
 ning resolutions.


 Artificially generated images can be used to exactly estimate the misclassi-
fication error of any    segmentation method.   For this   purpose,  expected
resulting images create1.pgm.gz (for noisytext1-6.pgm.gz) and bold.pgm.gz
for noisybold.pgm.gz, strips1.pgm.gz and strips2.pgm.gz) are included.  


List of real scanned images
---------------------------
1.  kamel1.pgm.gz
2.  kamel2.pgm.gz
3.  map.pgm.gz
4.  calender.pgm.gz
5.  blue_print.pgm.gz
6.  nbcheck.pgm.gz
7.  slant1.pgm.gz
8.  british_lib.pgm.gz
9.  test75.pgm.gz
10. test150.pgm.gz
11. test300.pgm.gz

List of artificially generated images
------------------------------------------

1.  noisytext1.pgm.gz
2.  noisytext3.pgm.gz
3.  noisytext4.pgm.gz
4.  noisytext5.pgm.gz
5.  noisytext6.pgm.gz
6.  noisybold.pgm.gz
7.  noisycircle.pgm.gz
8.  strips1.pgm.gz
9.  strips2.pgm.gz
10. creat1.pgm.gz
11. bold.pgm.gz


How to download
-----------------

ftp to  agora.leeds.ac.uk and go to /scs/ocr directory. You will find
testimages.readme and testimages.tar. Download them. Makesure that you are
using binary mode of transfer.



Useful References
----------------

M. Kamel and A. Zhao, Extraction of Binary Character/Graphics Images   from
Grayscale Document Images, CVGIP: Graphical Models and Image    Processing,
55:3, 203-217, 1993. 

L. Eikvil, T. Taxt and k. Moen, A Fast Adaptive Method for Binarization of
Document Images, Proc. of International Conference on Document Analysis and
recognition, France, 435-443, 1991.

N.B. Venkateswarlu, and R.D. Boyle, New Segmentation Techniques for Document
Image Analysis, Under revision for Image Vision and Computing, 1994. 




-- N.B. Venkateswarlu
   venkat@scs.leeds.ac.uk


Keywords: 

