X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: f996b,b639181160774d8d X-Google-Attributes: gidf996b,public X-Google-ArrivalTime: 2001-10-03 15:07:55 PST Path: archiver1.google.com!news1.google.com!sn-xit-02!supernews.com!newsfeed.direct.ca!look.ca!fu-berlin.de!uni-berlin.de!a1as08-p212.mch.tli.DE!not-for-mail From: Michael Schierl Newsgroups: alt.ascii-art Subject: Re: Nokia 8210 Date: Thu, 04 Oct 2001 00:08:17 +0200 Lines: 62 Message-ID: <9pg27j$i9jph$2@ID-39741.news.dfncis.de> References: <2GNs7.25353$Lc6.1427451@news1.cableinet.net> <9pamf7$h3u3l$4@ID-39741.news.dfncis.de> <3BB8EE51.38A2D080@rz.uni-karlsruhe.de> Reply-To: schierl@gmx.de NNTP-Posting-Host: a1as08-p212.mch.tli.de (195.252.164.212) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Trace: fu-berlin.de 1002146868 19189553 195.252.164.212 (16 [39741]) X-Newsreader: Forte Agent 1.8/32.548 Xref: archiver1.google.com alt.ascii-art:8327 Markus Gebhard wrote: > A few days ago I have written a simple algorithm to correct documents > with wrapped lines (this algorithm will be part of JavE 2.0). > Unfortunately it only works when some assumptions are true for the > wrapped image: > - there is only one line break in each line otherwise you have to guess. > - each (correct) line has a character different from [space] at the > first and last position Yes. That's my problem, too. Or - better the fact, that MSOE is cutting off those spaces. > - each line is wrapped that assumption only tells you that you can always combine two lines to one, so you know where the original line breaks were. a greedy algoritm and some heuristics could undergo that problem. > This is true for about 30% of the wrapped images I have seen in this > newsgroup. However it is not true for this nokia image. I guess there is > no algorithm that can fix this one. There had to be some heuristics... Like the algorithm to crack the TV channel "Premiere". Look what of the line is similar to another part. When I do it by hand, the lines should also not begin or end with spaces. I personally do it that way (by hand): 1. look for the longest line and create a reference line of that count of dashes. 2. add the line after that line and add dots to the reference line so that it is the same length. 3. Combine each line that is as long as the longest one (the dashes) with the last one. (Do not add spaces) 4. then start at the beginning of the image: a) If a line and the following two lines are shorter than the dashes&dots, combine them and add a some spaces (about ten) between them. b) If a line and the following line together are shorter than the dashes+dots, combine them and add spaces until it is the correct length. c) if 4 looks silly at a line and 5 looks better, undo 4 and do 5 instead. until you are at the end of the image. 5. Look at the lines that are too long (they were from 5) and change them so that it looks "good" I tried to program that once, but the last two points (4c and 5) are the problem. And, nevertheless, it did nor work with the Nokia picture. Michael