Text recognition in multimedia documents: a study of two neural-based OCRs using and avoiding character segmentation
Identifieur interne : 000121 ( Main/Exploration ); précédent : 000120; suivant : 000122Text recognition in multimedia documents: a study of two neural-based OCRs using and avoiding character segmentation
Auteurs : Khaoula Elagouni [France] ; Christophe Garcia [France] ; Franck Mamalet [France] ; Pascale Sebillot [France]Source :
- International journal on document analysis and recognition : (Print) [ 1433-2833 ] ; 2014.
Descripteurs français
- Pascal (Inist)
- Reconnaissance caractère, Texte, Reconnaissance forme, Analyse documentaire, Multimédia, Reconnaissance optique caractère, Accès contenu, Vision ordinateur, Linguistique mathématique, Signal vidéo, Analyse scène, Présentation document, Sémantique, Balayage, Sous titrage, Convolution, Réseau neuronal, Segmentation image, Scène naturelle.
- Wicri :
- topic : Multimédia.
English descriptors
- KwdEn :
- Caption, Character recognition, Computational linguistics, Computer vision, Content access, Convolution, Document analysis, Document layout, Image segmentation, Multimedia, Natural scenes, Neural network, Optical character recognition, Pattern recognition, Scanning, Scene analysis, Semantics, Text, Video signal.
Abstract
Text embedded in multimedia documents represents an important semantic information that helps to automatically access the content. This paper proposes two neural-based optical character recognition (OCR) systems that handle the text recognition problem in different ways. The first approach segments a text image into individual characters before recognizing them, while the second one avoids the segmentation step by integrating a multi-scale scanning scheme that allows to jointly localize and recognize characters at each position and scale. Some linguistic knowledge is also incorporated into the proposed schemes to remove errors due to recognition confusions. Both OCR systems are applied to caption texts embedded in videos and in natural scene images and provide outstanding results showing that the proposed approaches outperform the state-of-the-art methods.
Affiliations:
Links toward previous steps (curation, corpus...)
- to stream PascalFrancis, to step Corpus: 000009
- to stream PascalFrancis, to step Curation: 000756
- to stream PascalFrancis, to step Checkpoint: 000004
- to stream Main, to step Merge: 000122
- to stream Main, to step Curation: 000121
Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" level="a">Text recognition in multimedia documents: a study of two neural-based OCRs using and avoiding character segmentation</title>
<author><name sortKey="Elagouni, Khaoula" sort="Elagouni, Khaoula" uniqKey="Elagouni K" first="Khaoula" last="Elagouni">Khaoula Elagouni</name>
<affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Orange Labs R&D</s1>
<s2>35512 Cesson-Sévigné</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Région Bretagne</region>
<settlement type="city">Cesson-Sévigné</settlement>
</placeName>
</affiliation>
</author>
<author><name sortKey="Garcia, Christophe" sort="Garcia, Christophe" uniqKey="Garcia C" first="Christophe" last="Garcia">Christophe Garcia</name>
<affiliation wicri:level="3"><inist:fA14 i1="02"><s1>LIRIS/INSA de Lyon</s1>
<s2>69621 Villeurbanne</s2>
<s3>FRA</s3>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Auvergne-Rhône-Alpes</region>
<region type="old region" nuts="2">Rhône-Alpes</region>
<settlement type="city">Villeurbanne</settlement>
</placeName>
</affiliation>
</author>
<author><name sortKey="Mamalet, Franck" sort="Mamalet, Franck" uniqKey="Mamalet F" first="Franck" last="Mamalet">Franck Mamalet</name>
<affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Orange Labs R&D</s1>
<s2>35512 Cesson-Sévigné</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Région Bretagne</region>
<settlement type="city">Cesson-Sévigné</settlement>
</placeName>
</affiliation>
</author>
<author><name sortKey="Sebillot, Pascale" sort="Sebillot, Pascale" uniqKey="Sebillot P" first="Pascale" last="Sebillot">Pascale Sebillot</name>
<affiliation wicri:level="3"><inist:fA14 i1="03"><s1>IRISA/INSA de Rennes</s1>
<s2>35042 Rennes</s2>
<s3>FRA</s3>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Région Bretagne</region>
<settlement type="city">Rennes</settlement>
</placeName>
</affiliation>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">INIST</idno>
<idno type="inist">14-0199549</idno>
<date when="2014">2014</date>
<idno type="stanalyst">PASCAL 14-0199549 INIST</idno>
<idno type="RBID">Pascal:14-0199549</idno>
<idno type="wicri:Area/PascalFrancis/Corpus">000009</idno>
<idno type="wicri:Area/PascalFrancis/Curation">000756</idno>
<idno type="wicri:Area/PascalFrancis/Checkpoint">000004</idno>
<idno type="wicri:doubleKey">1433-2833:2014:Elagouni K:text:recognition:in</idno>
<idno type="wicri:Area/Main/Merge">000122</idno>
<idno type="wicri:Area/Main/Curation">000121</idno>
<idno type="wicri:Area/Main/Exploration">000121</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en" level="a">Text recognition in multimedia documents: a study of two neural-based OCRs using and avoiding character segmentation</title>
<author><name sortKey="Elagouni, Khaoula" sort="Elagouni, Khaoula" uniqKey="Elagouni K" first="Khaoula" last="Elagouni">Khaoula Elagouni</name>
<affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Orange Labs R&D</s1>
<s2>35512 Cesson-Sévigné</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Région Bretagne</region>
<settlement type="city">Cesson-Sévigné</settlement>
</placeName>
</affiliation>
</author>
<author><name sortKey="Garcia, Christophe" sort="Garcia, Christophe" uniqKey="Garcia C" first="Christophe" last="Garcia">Christophe Garcia</name>
<affiliation wicri:level="3"><inist:fA14 i1="02"><s1>LIRIS/INSA de Lyon</s1>
<s2>69621 Villeurbanne</s2>
<s3>FRA</s3>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Auvergne-Rhône-Alpes</region>
<region type="old region" nuts="2">Rhône-Alpes</region>
<settlement type="city">Villeurbanne</settlement>
</placeName>
</affiliation>
</author>
<author><name sortKey="Mamalet, Franck" sort="Mamalet, Franck" uniqKey="Mamalet F" first="Franck" last="Mamalet">Franck Mamalet</name>
<affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Orange Labs R&D</s1>
<s2>35512 Cesson-Sévigné</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Région Bretagne</region>
<settlement type="city">Cesson-Sévigné</settlement>
</placeName>
</affiliation>
</author>
<author><name sortKey="Sebillot, Pascale" sort="Sebillot, Pascale" uniqKey="Sebillot P" first="Pascale" last="Sebillot">Pascale Sebillot</name>
<affiliation wicri:level="3"><inist:fA14 i1="03"><s1>IRISA/INSA de Rennes</s1>
<s2>35042 Rennes</s2>
<s3>FRA</s3>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Région Bretagne</region>
<settlement type="city">Rennes</settlement>
</placeName>
</affiliation>
</author>
</analytic>
<series><title level="j" type="main">International journal on document analysis and recognition : (Print)</title>
<title level="j" type="abbreviated">Int. j. doc. anal. recognit. : (Print)</title>
<idno type="ISSN">1433-2833</idno>
<imprint><date when="2014">2014</date>
</imprint>
</series>
</biblStruct>
</sourceDesc>
<seriesStmt><title level="j" type="main">International journal on document analysis and recognition : (Print)</title>
<title level="j" type="abbreviated">Int. j. doc. anal. recognit. : (Print)</title>
<idno type="ISSN">1433-2833</idno>
</seriesStmt>
</fileDesc>
<profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Caption</term>
<term>Character recognition</term>
<term>Computational linguistics</term>
<term>Computer vision</term>
<term>Content access</term>
<term>Convolution</term>
<term>Document analysis</term>
<term>Document layout</term>
<term>Image segmentation</term>
<term>Multimedia</term>
<term>Natural scenes</term>
<term>Neural network</term>
<term>Optical character recognition</term>
<term>Pattern recognition</term>
<term>Scanning</term>
<term>Scene analysis</term>
<term>Semantics</term>
<term>Text</term>
<term>Video signal</term>
</keywords>
<keywords scheme="Pascal" xml:lang="fr"><term>Reconnaissance caractère</term>
<term>Texte</term>
<term>Reconnaissance forme</term>
<term>Analyse documentaire</term>
<term>Multimédia</term>
<term>Reconnaissance optique caractère</term>
<term>Accès contenu</term>
<term>Vision ordinateur</term>
<term>Linguistique mathématique</term>
<term>Signal vidéo</term>
<term>Analyse scène</term>
<term>Présentation document</term>
<term>Sémantique</term>
<term>Balayage</term>
<term>Sous titrage</term>
<term>Convolution</term>
<term>Réseau neuronal</term>
<term>Segmentation image</term>
<term>Scène naturelle</term>
</keywords>
<keywords scheme="Wicri" type="topic" xml:lang="fr"><term>Multimédia</term>
</keywords>
</textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">Text embedded in multimedia documents represents an important semantic information that helps to automatically access the content. This paper proposes two neural-based optical character recognition (OCR) systems that handle the text recognition problem in different ways. The first approach segments a text image into individual characters before recognizing them, while the second one avoids the segmentation step by integrating a multi-scale scanning scheme that allows to jointly localize and recognize characters at each position and scale. Some linguistic knowledge is also incorporated into the proposed schemes to remove errors due to recognition confusions. Both OCR systems are applied to caption texts embedded in videos and in natural scene images and provide outstanding results showing that the proposed approaches outperform the state-of-the-art methods.</div>
</front>
</TEI>
<affiliations><list><country><li>France</li>
</country>
<region><li>Auvergne-Rhône-Alpes</li>
<li>Rhône-Alpes</li>
<li>Région Bretagne</li>
</region>
<settlement><li>Cesson-Sévigné</li>
<li>Rennes</li>
<li>Villeurbanne</li>
</settlement>
</list>
<tree><country name="France"><region name="Région Bretagne"><name sortKey="Elagouni, Khaoula" sort="Elagouni, Khaoula" uniqKey="Elagouni K" first="Khaoula" last="Elagouni">Khaoula Elagouni</name>
</region>
<name sortKey="Garcia, Christophe" sort="Garcia, Christophe" uniqKey="Garcia C" first="Christophe" last="Garcia">Christophe Garcia</name>
<name sortKey="Mamalet, Franck" sort="Mamalet, Franck" uniqKey="Mamalet F" first="Franck" last="Mamalet">Franck Mamalet</name>
<name sortKey="Sebillot, Pascale" sort="Sebillot, Pascale" uniqKey="Sebillot P" first="Pascale" last="Sebillot">Pascale Sebillot</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 000121 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 000121 | 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:14-0199549 |texte= Text recognition in multimedia documents: a study of two neural-based OCRs using and avoiding character segmentation }}
This area was generated with Dilib version V0.6.32. |