Recognition of Multifont Text Using Markov Models
Identifieur interne : 000379 ( France/Analysis ); précédent : 000378; suivant : 000380Recognition of Multifont Text Using Markov Models
Auteurs : J.-C. Anigbogu ; Abdel Belaïd [France]Source :
English descriptors
- KwdEn :
Abstract
This paper exposes a stochastic modeling approach to multifont printed text using Hidden Markov Models. The system uses stable features like profiles, global character shape, presence and position of holes, etc. to capture as much as possible the variations in character shapes in different fonts. We first determine the identity of the predominant font, paragraph by paragraph so as to limit the number of models to deal with. This is achieved by projection some features into hyperspace and using Euclidian distance measures to determine proximity to font prototypes constructed during a learning phase. Given that HMM presents a global view of the forms, deterministic decision-trees are used to channel the system towards appropriate models. We also use such heuristics as presence of ascenders and descenders to construct these trees.
Affiliations:
- France
- Alsace-Champagne-Ardenne-Lorraine, Région Lorraine
- Nancy
- Centre national de la recherche scientifique, Institut national de recherche en informatique et en automatique, Laboratoire lorrain de recherche en informatique et ses applications, Université de Lorraine
Links toward previous steps (curation, corpus...)
- to stream Crin, to step Corpus: 000007
- to stream Crin, to step Curation: 000007
- to stream Crin, to step Checkpoint: 000006
- to stream Main, to step Merge: 003594
- to stream Main, to step Curation: 003407
- to stream Main, to step Exploration: 003407
- to stream France, to step Extraction: 000379
Links to Exploration step
CRIN:anigbogu91aLe document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" wicri:score="371">Recognition of Multifont Text Using Markov Models</title></titleStmt><publicationStmt><idno type="RBID">CRIN:anigbogu91a</idno><date when="1991" year="1991">1991</date><idno type="wicri:Area/Crin/Corpus">000007</idno><idno type="wicri:Area/Crin/Curation">000007</idno><idno type="wicri:Area/Crin/Checkpoint">000006</idno><idno type="wicri:Area/Main/Merge">003594</idno><idno type="wicri:Area/Main/Curation">003407</idno><idno type="wicri:Area/Main/Exploration">003407</idno><idno type="wicri:Area/France/Extraction">000379</idno></publicationStmt><sourceDesc><biblStruct><analytic><title xml:lang="en">Recognition of Multifont Text Using Markov Models</title><author><name sortKey="Anigbogu, J C" sort="Anigbogu, J C" uniqKey="Anigbogu J" first="J.-C." last="Anigbogu">J.-C. Anigbogu</name></author><author><name sortKey="Belaid, A" sort="Belaid, A" uniqKey="Belaid A" first="A." last="Belaïd">Abdel Belaïd</name><affiliation><country>France</country><placeName><settlement type="city">Nancy</settlement><region type="region" nuts="2">Alsace-Champagne-Ardenne-Lorraine</region><region type="region" nuts="2">Région Lorraine</region></placeName><orgName type="laboratoire" n="5">Laboratoire lorrain de recherche en informatique et ses applications</orgName><orgName type="university">Université de Lorraine</orgName><orgName type="institution">Centre national de la recherche scientifique</orgName><orgName type="institution">Institut national de recherche en informatique et en automatique</orgName></affiliation></author></analytic></biblStruct></sourceDesc></fileDesc><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Markov model</term><term>OCR</term><term>character recognition</term></keywords></textClass></profileDesc></teiHeader><front><div type="abstract" xml:lang="en" wicri:score="632">This paper exposes a stochastic modeling approach to multifont printed text using Hidden Markov Models. The system uses stable features like profiles, global character shape, presence and position of holes, etc. to capture as much as possible the variations in character shapes in different fonts. We first determine the identity of the predominant font, paragraph by paragraph so as to limit the number of models to deal with. This is achieved by projection some features into hyperspace and using Euclidian distance measures to determine proximity to font prototypes constructed during a learning phase. Given that HMM presents a global view of the forms, deterministic decision-trees are used to channel the system towards appropriate models. We also use such heuristics as presence of ascenders and descenders to construct these trees.</div></front></TEI><affiliations><list><country><li>France</li></country><region><li>Alsace-Champagne-Ardenne-Lorraine</li><li>Région Lorraine</li></region><settlement><li>Nancy</li></settlement><orgName><li>Centre national de la recherche scientifique</li><li>Institut national de recherche en informatique et en automatique</li><li>Laboratoire lorrain de recherche en informatique et ses applications</li><li>Université de Lorraine</li></orgName></list><tree><noCountry><name sortKey="Anigbogu, J C" sort="Anigbogu, J C" uniqKey="Anigbogu J" first="J.-C." last="Anigbogu">J.-C. Anigbogu</name></noCountry><country name="France"><region name="Alsace-Champagne-Ardenne-Lorraine"><name sortKey="Belaid, A" sort="Belaid, A" uniqKey="Belaid A" first="A." last="Belaïd">Abdel Belaïd</name></region></country></tree></affiliations></record>Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Ticri/CIDE/explor/OcrV1/Data/France/Analysis
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 000379 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/France/Analysis/biblio.hfd -nk 000379 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Ticri/CIDE
|area= OcrV1
|flux= France
|étape= Analysis
|type= RBID
|clé= CRIN:anigbogu91a
|texte= Recognition of Multifont Text Using Markov Models
}}
| This area was generated with Dilib version V0.6.32. |  |