Partitioning of the degradation space for OCR training
Identifieur interne : 001158 ( Main/Exploration ); précédent : 001157; suivant : 001159Partitioning of the degradation space for OCR training
Auteurs : Elisa H. Barney Smith [États-Unis] ; Tim Andersen [États-Unis]Source :
- Proceedings of SPIE, the International Society for Optical Engineering [ 0277-786X ] ; 2006.
Descripteurs français
- Pascal (Inist)
English descriptors
- KwdEn :
Abstract
Generally speaking optical character recognition algorithms tend to perform better when presented with homogeneous data. This paper studies a method that is designed to increase the homogeneity of training data, based on an understanding of the types of degradations that occur during the printing and scanning process, and how these degradations affect the homogeneity of the data. While it has been shown that dividing the degradation space by edge spread improves recognition accuracy over dividing the degradation space by threshold or point spread function width alone, the challenge is in deciding how many partitions and at what value of edge spread the divisions should be made. Clustering of different types of character features, fonts, sizes, resolutions and noise levels shows that edge spread is indeed shown to be a strong indicator of the homogeneity of character data clusters.
Affiliations:
Links toward previous steps (curation, corpus...)
- to stream PascalFrancis, to step Corpus: 000334
- to stream PascalFrancis, to step Curation: 000452
- to stream PascalFrancis, to step Checkpoint: 000322
- to stream Main, to step Merge: 001183
- to stream Main, to step Curation: 001158
Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" level="a">Partitioning of the degradation space for OCR training</title>
<author><name sortKey="Smith, Elisa H Barney" sort="Smith, Elisa H Barney" uniqKey="Smith E" first="Elisa H. Barney" last="Smith">Elisa H. Barney Smith</name>
<affiliation wicri:level="2"><inist:fA14 i1="01"><s1>Boise State University</s1>
<s2>Boise, ID</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>États-Unis</country>
<placeName><region type="state">Idaho</region>
</placeName>
</affiliation>
</author>
<author><name sortKey="Andersen, Tim" sort="Andersen, Tim" uniqKey="Andersen T" first="Tim" last="Andersen">Tim Andersen</name>
<affiliation wicri:level="2"><inist:fA14 i1="01"><s1>Boise State University</s1>
<s2>Boise, ID</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>États-Unis</country>
<placeName><region type="state">Idaho</region>
</placeName>
</affiliation>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">INIST</idno>
<idno type="inist">07-0377967</idno>
<date when="2006">2006</date>
<idno type="stanalyst">PASCAL 07-0377967 INIST</idno>
<idno type="RBID">Pascal:07-0377967</idno>
<idno type="wicri:Area/PascalFrancis/Corpus">000334</idno>
<idno type="wicri:Area/PascalFrancis/Curation">000452</idno>
<idno type="wicri:Area/PascalFrancis/Checkpoint">000322</idno>
<idno type="wicri:doubleKey">0277-786X:2006:Smith E:partitioning:of:the</idno>
<idno type="wicri:Area/Main/Merge">001183</idno>
<idno type="wicri:Area/Main/Curation">001158</idno>
<idno type="wicri:Area/Main/Exploration">001158</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en" level="a">Partitioning of the degradation space for OCR training</title>
<author><name sortKey="Smith, Elisa H Barney" sort="Smith, Elisa H Barney" uniqKey="Smith E" first="Elisa H. Barney" last="Smith">Elisa H. Barney Smith</name>
<affiliation wicri:level="2"><inist:fA14 i1="01"><s1>Boise State University</s1>
<s2>Boise, ID</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>États-Unis</country>
<placeName><region type="state">Idaho</region>
</placeName>
</affiliation>
</author>
<author><name sortKey="Andersen, Tim" sort="Andersen, Tim" uniqKey="Andersen T" first="Tim" last="Andersen">Tim Andersen</name>
<affiliation wicri:level="2"><inist:fA14 i1="01"><s1>Boise State University</s1>
<s2>Boise, ID</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>États-Unis</country>
<placeName><region type="state">Idaho</region>
</placeName>
</affiliation>
</author>
</analytic>
<series><title level="j" type="main">Proceedings of SPIE, the International Society for Optical Engineering</title>
<idno type="ISSN">0277-786X</idno>
<imprint><date when="2006">2006</date>
</imprint>
</series>
</biblStruct>
</sourceDesc>
<seriesStmt><title level="j" type="main">Proceedings of SPIE, the International Society for Optical Engineering</title>
<idno type="ISSN">0277-786X</idno>
</seriesStmt>
</fileDesc>
<profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Accuracy</term>
<term>Algorithm</term>
<term>Automatic classification</term>
<term>Degradation</term>
<term>Learning</term>
<term>Optical character recognition</term>
<term>Partition method</term>
<term>Point spread function</term>
</keywords>
<keywords scheme="Pascal" xml:lang="fr"><term>Algorithme</term>
<term>Fonction étalement point</term>
<term>Méthode partition</term>
<term>Dégradation</term>
<term>Reconnaissance optique caractère</term>
<term>Apprentissage</term>
<term>Précision</term>
<term>Classification automatique</term>
<term>4230S</term>
</keywords>
</textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">Generally speaking optical character recognition algorithms tend to perform better when presented with homogeneous data. This paper studies a method that is designed to increase the homogeneity of training data, based on an understanding of the types of degradations that occur during the printing and scanning process, and how these degradations affect the homogeneity of the data. While it has been shown that dividing the degradation space by edge spread improves recognition accuracy over dividing the degradation space by threshold or point spread function width alone, the challenge is in deciding how many partitions and at what value of edge spread the divisions should be made. Clustering of different types of character features, fonts, sizes, resolutions and noise levels shows that edge spread is indeed shown to be a strong indicator of the homogeneity of character data clusters.</div>
</front>
</TEI>
<affiliations><list><country><li>États-Unis</li>
</country>
<region><li>Idaho</li>
</region>
</list>
<tree><country name="États-Unis"><region name="Idaho"><name sortKey="Smith, Elisa H Barney" sort="Smith, Elisa H Barney" uniqKey="Smith E" first="Elisa H. Barney" last="Smith">Elisa H. Barney Smith</name>
</region>
<name sortKey="Andersen, Tim" sort="Andersen, Tim" uniqKey="Andersen T" first="Tim" last="Andersen">Tim Andersen</name>
</country>
</tree>
</affiliations>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Ticri/CIDE/explor/OcrV1/Data/Main/Exploration
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 001158 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 001158 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Ticri/CIDE |area= OcrV1 |flux= Main |étape= Exploration |type= RBID |clé= Pascal:07-0377967 |texte= Partitioning of the degradation space for OCR training }}
This area was generated with Dilib version V0.6.32. |