Xref: utzoo comp.windows.x:11684 comp.graphics:6550
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!bloom-beacon!mit-eddie!eric
From: eric@eddie.MIT.EDU (Eric Van Tassell)
Newsgroups: comp.windows.x,comp.graphics
Subject: A fast algorithm to find all closed regions in an image
Keywords: closed regions
Message-ID: <12247@eddie.MIT.EDU>
Date: 20 Jul 89 23:29:29 GMT
Organization: MIT EE/CS Computer Facility, Cambridge, MA
Lines: 41


Does anybody have a fast algorithm for taking an image (collection of pixel
values) and determining the closed regions? See ascii pic below.



|------------------------------------------------------------|
| *             *                                            |
|  *     1      *                                            |
|   *           *                                            |
|    ****************                                        |
|             *      *                                       |
|            *   2    *                                      |
|            *      ***********                              |
|            *     *    *   3  *                             |
|            *******  **********                             |
|                    **   *                                  |
|        ************      *******                           |
|       *              4          *                          |
|       ***************************                          |
|                                                            |
|                                                            |
|------------------------------------------------------------|


     What I want is an algorithm (fast) that will tell me that regions
     2,3,4 are closed (and can be filled) but region 1 is not. By "telling me
     that they are closed I supposed I mean that the algorithm should return
     a list of lists of points representing the polygonal boundaries of
     all enclosed regions. I will be doing this under X11R3 but I am mostly
     concerned with a fast algorithm and dont really care if it is X specific.
     As a matter of fact since I will have to port this to PM, dependance on
     X would be a lose. But since I am asking for free advice I'll take
     anything. TIA!

eric@eddie.mit.edu
-- 
Eric Van Tassell(dlcdev!eric@eddie.mit.edu)
Progress Software Corp.
22B Cotton Rd.
Nashua, NH 03063 {603-882-2488}