当前位置: 代码迷 >> 共享软件 >> :翻译下文,能翻多少是多少(2篇文章)
  详细解决方案

:翻译下文,能翻多少是多少(2篇文章)

热度:7750   发布时间:2013-02-26 00:00:00.0
求助:翻译下文,能翻多少是多少(2篇文章)
Segmentation     of   page   images   having   artifacts   of   photocopying   and   scanning  
  L.   Cinque   ,     ,   a,   S.   Levialdia,   L.   Lombardib   and   S.   Tanimotoc  
a   Dipartimento   di   Scienze   dell 'Informazione,   University   of   Rome   “La   Sapienza”,   Via   Salaria   113,   00198   Rome,   Italy
b   Dipartimento   di   Informatica   e   Sistemistica,   University   of   Pavia,   Via   Ferrata   1,   27100   Pavia,   Italy
c   Department   of   Computer   Sci.   and   Engineering,   Box   352350,   University   of   Washington,   Seattle,   WA   98195,   USA  
Received   18   June   1999;   revised   16   February   2001;   accepted   26   February   2001   Available   online   11   February   2002.  


Abstract  
The   analysis   of   scanned   documents   is   important   in   the   construction   of   digital   libraries   and   paperless   offices.   One   significant   challenge   is   coping   with   artifacts   of   photocopying   and   scanning.   We   present   a   series   of   simple   techniques   for   handling   these   difficulties.   Using   125   images   of   the   University   of   Washington   scanned   documents   database,   we   demonstrate   the   effectiveness   of   these   methods   in   preparing   the   images   for     segmentation     by   a   multiresolution   algorithm.  
Author   Keywords:   Document   analysis;   Artifact   elimination;     Segmentation   ;   Print-through;   Marginal   artifact;   Partial   extra   page;   Digital   library  


Article   Outline  
1.   Introduction  
1.1.   General   motivation  
1.2.   Problem   description
2.   Previous   work  
3.   Processing   methods  
3.1.   Eliminating   print-through
Algorithm   1—[Treatment   of   print-through]  
3.2.   Marginal   artifacts   and   partial   extra   pages
Algorithm   2—[Treatment   of   marginal   artifacts   and   partial   extra   pages]  
4.   Multiresolution   segmentation   method  
4.1.   Phase   1:   construction   of   feature   pyramids  
4.2.   Phase   2:   classification   of   regions
5.   Experimental   results   and   discussion  
5.1.   Computational   considerations
6.   Conclusions  
References  
Vitae


1.   Introduction  
1.1.   General   motivation  
Online   digital   libraries   can   provide   improved   distribution   of   information   and   more   flexible   access   via   search   algorithms   than   can   traditional   print   libraries.   However,   adding   existing   print   materials   to   electronic   libraries   is   a   costly,   slow   process   unless   good   automated   procedures   can   be   developed.   After   academic   journal   articles   have   been   photocopied   and/or   scanned   from   their   bound,   printed   versions,   various   artifacts   have   often   been   introduced   into   the   images   that   make   further   analysis   difficult.   Either   these   artifacts   need   to   be   removed   before   further   processing,   or   special   considerations   must   be   given   to   the   following   processing   steps   to   make   them   tolerant   of   the   artifacts.   We   addressed   the   problem   of   artifacts   by   developing   means   to   reduce   and/or   eliminate   them   from   the   scanned   document   images   prior   to   segmentation.  
  相关解决方案