SnooperText: A text detection system for automatic indexing of urban scenes
Identifieur interne : 000123 ( Main/Exploration ); précédent : 000122; suivant : 000124SnooperText: A text detection system for automatic indexing of urban scenes
Auteurs : Rodrigo Minetto [Brésil] ; Nicolas Thome [France] ; Matthieu Cord [France] ; Neucimar J. Leite [Brésil] ; Jorge Stolfi [Brésil]Source :
- Computer vision and image understanding : (Print) [ 1077-3142 ] ; 2014.
Descripteurs français
- Pascal (Inist)
- Recherche information, Texte, Mesure automatique, Donnée textuelle, Vision ordinateur, Traitement image, Classification, Forme géométrique, Analyse contenu, Histogramme, Image multiple, Extraction information, Métrique, Filtre morphologique, Extraction forme, Reconnaissance caractère, ., Segmentation image, Détection objet.
- Wicri :
- topic : Classification.
English descriptors
- KwdEn :
- Automatic measurement, Character recognition, Classification, Computer vision, Content analysis, Geometrical shape, Histogram, Image processing, Image segmentation, Information extraction, Information retrieval, Metric, Morphological filter, Multiple image, Object detection, Pattern extraction, Text, Textual data.
Abstract
We describe SNOOPERTEXT, an original detector for textual information embedded in photos of building façades (such as names of stores, products and services) that we developed for the iTowns urban geographic information project. SNOOPERTEXT locates candidate characters by using toggle-mapping image segmentation and character/non-character classification based on shape descriptors. The candidate characters are then grouped to form either candidate words or candidate text lines. These candidate regions are then validated by a text/non-text classifier using a HOG-based descriptor specifically tuned to single-line text regions. These operations are applied at multiple image scales in order to suppress irrelevant detail in character shapes and to avoid the use of overly large kernels in the segmentation. We show that SNOOPERTEXT outperforms other published state-of-the-art text detection algorithms on standard image benchmarks. We also describe two metrics to evaluate the end-to-end performance of text extraction systems, and show that the use of SNOOPERTEXT as a pre-filter significantly improves the performance of a general-purpose OCR algorithm when applied to photos of urban scenes.
Url:
Affiliations:
Links toward previous steps (curation, corpus...)
- to stream PascalFrancis, to step Corpus: 000015
- to stream PascalFrancis, to step Curation: 000749
- to stream PascalFrancis, to step Checkpoint: 000006
- to stream Main, to step Merge: 000124
- to stream Hal, to step Corpus: 000110
- to stream Hal, to step Curation: 000110
- to stream Hal, to step Checkpoint: 000032
- to stream Main, to step Merge: 000072
- to stream Main, to step Curation: 000123
Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" level="a">SnooperText: A text detection system for automatic indexing of urban scenes</title><author><name sortKey="Minetto, Rodrigo" sort="Minetto, Rodrigo" uniqKey="Minetto R" first="Rodrigo" last="Minetto">Rodrigo Minetto</name><affiliation wicri:level="1"><inist:fA14 i1="01"><s1>DAINF, Federal University of Technology</s1><s2>Paraná, Curitiba</s2><s3>BRA</s3><sZ>1 aut.</sZ></inist:fA14><country>Brésil</country><wicri:noRegion>Paraná, Curitiba</wicri:noRegion></affiliation></author><author><name sortKey="Thome, Nicolas" sort="Thome, Nicolas" uniqKey="Thome N" first="Nicolas" last="Thome">Nicolas Thome</name><affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Laboratoire d'Informatique Paris 6 (LIP6), Université Pierre et Marie Curie</s1><s2>Paris</s2><s3>FRA</s3><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><settlement type="city">Paris</settlement></placeName><orgName type="university">Université Pierre-et-Marie-Curie</orgName><placeName><settlement type="city">Paris</settlement><region type="region" nuts="2">Île-de-France</region></placeName></affiliation></author><author><name sortKey="Cord, Matthieu" sort="Cord, Matthieu" uniqKey="Cord M" first="Matthieu" last="Cord">Matthieu Cord</name><affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Laboratoire d'Informatique Paris 6 (LIP6), Université Pierre et Marie Curie</s1><s2>Paris</s2><s3>FRA</s3><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><settlement type="city">Paris</settlement></placeName><orgName type="university">Université Pierre-et-Marie-Curie</orgName><placeName><settlement type="city">Paris</settlement><region type="region" nuts="2">Île-de-France</region></placeName></affiliation></author><author><name sortKey="Leite, Neucimar J" sort="Leite, Neucimar J" uniqKey="Leite N" first="Neucimar J." last="Leite">Neucimar J. Leite</name><affiliation wicri:level="1"><inist:fA14 i1="03"><s1>Institute of Computing, University of Campinas</s1><s2>Campinas</s2><s3>BRA</s3><sZ>4 aut.</sZ><sZ>5 aut.</sZ></inist:fA14><country>Brésil</country><wicri:noRegion>Campinas</wicri:noRegion></affiliation></author><author><name sortKey="Stolfi, Jorge" sort="Stolfi, Jorge" uniqKey="Stolfi J" first="Jorge" last="Stolfi">Jorge Stolfi</name><affiliation wicri:level="1"><inist:fA14 i1="03"><s1>Institute of Computing, University of Campinas</s1><s2>Campinas</s2><s3>BRA</s3><sZ>4 aut.</sZ><sZ>5 aut.</sZ></inist:fA14><country>Brésil</country><wicri:noRegion>Campinas</wicri:noRegion></affiliation></author></titleStmt><publicationStmt><idno type="wicri:source">INIST</idno><idno type="inist">14-0142487</idno><date when="2014">2014</date><idno type="stanalyst">PASCAL 14-0142487 INIST</idno><idno type="RBID">Pascal:14-0142487</idno><idno type="wicri:Area/PascalFrancis/Corpus">000015</idno><idno type="wicri:Area/PascalFrancis/Curation">000749</idno><idno type="wicri:Area/PascalFrancis/Checkpoint">000006</idno><idno type="wicri:doubleKey">1077-3142:2014:Minetto R:snoopertext:a:text</idno><idno type="wicri:Area/Main/Merge">000124</idno><idno type="wicri:source">HAL</idno><idno type="RBID">Hal:hal-01185469</idno><idno type="url">https://hal.archives-ouvertes.fr/hal-01185469</idno><idno type="wicri:Area/Hal/Corpus">000110</idno><idno type="wicri:Area/Hal/Curation">000110</idno><idno type="wicri:Area/Hal/Checkpoint">000032</idno><idno type="wicri:doubleKey">1077-3142:2014:Minetto R:snoopertext:a:text</idno><idno type="wicri:Area/Main/Merge">000072</idno><idno type="wicri:Area/Main/Curation">000123</idno><idno type="wicri:Area/Main/Exploration">000123</idno></publicationStmt><sourceDesc><biblStruct><analytic><title xml:lang="en" level="a">SnooperText: A text detection system for automatic indexing of urban scenes</title><author><name sortKey="Minetto, Rodrigo" sort="Minetto, Rodrigo" uniqKey="Minetto R" first="Rodrigo" last="Minetto">Rodrigo Minetto</name><affiliation wicri:level="1"><inist:fA14 i1="01"><s1>DAINF, Federal University of Technology</s1><s2>Paraná, Curitiba</s2><s3>BRA</s3><sZ>1 aut.</sZ></inist:fA14><country>Brésil</country><wicri:noRegion>Paraná, Curitiba</wicri:noRegion></affiliation></author><author><name sortKey="Thome, Nicolas" sort="Thome, Nicolas" uniqKey="Thome N" first="Nicolas" last="Thome">Nicolas Thome</name><affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Laboratoire d'Informatique Paris 6 (LIP6), Université Pierre et Marie Curie</s1><s2>Paris</s2><s3>FRA</s3><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><settlement type="city">Paris</settlement></placeName><orgName type="university">Université Pierre-et-Marie-Curie</orgName><placeName><settlement type="city">Paris</settlement><region type="region" nuts="2">Île-de-France</region></placeName></affiliation></author><author><name sortKey="Cord, Matthieu" sort="Cord, Matthieu" uniqKey="Cord M" first="Matthieu" last="Cord">Matthieu Cord</name><affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Laboratoire d'Informatique Paris 6 (LIP6), Université Pierre et Marie Curie</s1><s2>Paris</s2><s3>FRA</s3><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><settlement type="city">Paris</settlement></placeName><orgName type="university">Université Pierre-et-Marie-Curie</orgName><placeName><settlement type="city">Paris</settlement><region type="region" nuts="2">Île-de-France</region></placeName></affiliation></author><author><name sortKey="Leite, Neucimar J" sort="Leite, Neucimar J" uniqKey="Leite N" first="Neucimar J." last="Leite">Neucimar J. Leite</name><affiliation wicri:level="1"><inist:fA14 i1="03"><s1>Institute of Computing, University of Campinas</s1><s2>Campinas</s2><s3>BRA</s3><sZ>4 aut.</sZ><sZ>5 aut.</sZ></inist:fA14><country>Brésil</country><wicri:noRegion>Campinas</wicri:noRegion></affiliation></author><author><name sortKey="Stolfi, Jorge" sort="Stolfi, Jorge" uniqKey="Stolfi J" first="Jorge" last="Stolfi">Jorge Stolfi</name><affiliation wicri:level="1"><inist:fA14 i1="03"><s1>Institute of Computing, University of Campinas</s1><s2>Campinas</s2><s3>BRA</s3><sZ>4 aut.</sZ><sZ>5 aut.</sZ></inist:fA14><country>Brésil</country><wicri:noRegion>Campinas</wicri:noRegion></affiliation></author></analytic><series><title level="j" type="main">Computer vision and image understanding : (Print)</title><title level="j" type="abbreviated">Comput. vis. image underst. : (Print)</title><idno type="ISSN">1077-3142</idno><imprint><date when="2014">2014</date></imprint></series></biblStruct></sourceDesc><seriesStmt><title level="j" type="main">Computer vision and image understanding : (Print)</title><title level="j" type="abbreviated">Comput. vis. image underst. : (Print)</title><idno type="ISSN">1077-3142</idno></seriesStmt></fileDesc><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Automatic measurement</term><term>Character recognition</term><term>Classification</term><term>Computer vision</term><term>Content analysis</term><term>Geometrical shape</term><term>Histogram</term><term>Image processing</term><term>Image segmentation</term><term>Information extraction</term><term>Information retrieval</term><term>Metric</term><term>Morphological filter</term><term>Multiple image</term><term>Object detection</term><term>Pattern extraction</term><term>Text</term><term>Textual data</term></keywords><keywords scheme="Pascal" xml:lang="fr"><term>Recherche information</term><term>Texte</term><term>Mesure automatique</term><term>Donnée textuelle</term><term>Vision ordinateur</term><term>Traitement image</term><term>Classification</term><term>Forme géométrique</term><term>Analyse contenu</term><term>Histogramme</term><term>Image multiple</term><term>Extraction information</term><term>Métrique</term><term>Filtre morphologique</term><term>Extraction forme</term><term>Reconnaissance caractère</term><term>.</term><term>Segmentation image</term><term>Détection objet</term></keywords><keywords scheme="Wicri" type="topic" xml:lang="fr"><term>Classification</term></keywords></textClass></profileDesc></teiHeader><front><div type="abstract" xml:lang="en">We describe SNOOPERTEXT, an original detector for textual information embedded in photos of building façades (such as names of stores, products and services) that we developed for the iTowns urban geographic information project. SNOOPERTEXT locates candidate characters by using toggle-mapping image segmentation and character/non-character classification based on shape descriptors. The candidate characters are then grouped to form either candidate words or candidate text lines. These candidate regions are then validated by a text/non-text classifier using a HOG-based descriptor specifically tuned to single-line text regions. These operations are applied at multiple image scales in order to suppress irrelevant detail in character shapes and to avoid the use of overly large kernels in the segmentation. We show that SNOOPERTEXT outperforms other published state-of-the-art text detection algorithms on standard image benchmarks. We also describe two metrics to evaluate the end-to-end performance of text extraction systems, and show that the use of SNOOPERTEXT as a pre-filter significantly improves the performance of a general-purpose OCR algorithm when applied to photos of urban scenes.</div></front></TEI><affiliations><list><country><li>Brésil</li><li>France</li></country><region><li>Île-de-France</li></region><settlement><li>Paris</li></settlement><orgName><li>Université Pierre-et-Marie-Curie</li></orgName></list><tree><country name="Brésil"><noRegion><name sortKey="Minetto, Rodrigo" sort="Minetto, Rodrigo" uniqKey="Minetto R" first="Rodrigo" last="Minetto">Rodrigo Minetto</name></noRegion><name sortKey="Leite, Neucimar J" sort="Leite, Neucimar J" uniqKey="Leite N" first="Neucimar J." last="Leite">Neucimar J. Leite</name><name sortKey="Stolfi, Jorge" sort="Stolfi, Jorge" uniqKey="Stolfi J" first="Jorge" last="Stolfi">Jorge Stolfi</name></country><country name="France"><noRegion><name sortKey="Thome, Nicolas" sort="Thome, Nicolas" uniqKey="Thome N" first="Nicolas" last="Thome">Nicolas Thome</name></noRegion><name sortKey="Cord, Matthieu" sort="Cord, Matthieu" uniqKey="Cord M" first="Matthieu" last="Cord">Matthieu Cord</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 000123 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 000123 | 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:14-0142487
|texte= SnooperText: A text detection system for automatic indexing of urban scenes
}}
| This area was generated with Dilib version V0.6.32. |  |