Feature extraction by best anisotropic Haar bases in an OCR system
Identifieur interne : 001633 ( Main/Curation ); précédent : 001632; suivant : 001634Feature extraction by best anisotropic Haar bases in an OCR system
Auteurs : Atanas Gotchev [Finlande] ; Dmytro Rusanovskyy [Finlande] ; Roumen Popov [Royaume-Uni] ; Karen Egiazarian [Finlande] ; Jaakko Astola [Finlande]Source :
- SPIE proceedings series [ 1017-2653 ] ; 2004.
Descripteurs français
- Pascal (Inist)
- Wicri :
- topic : Télévision, Base de données.
English descriptors
- KwdEn :
Abstract
In this contribution, we explore the best basis paradigm for in feature extraction. According to this paradigm, a library of bases is built and the best basis is found for a given signal class with respect to some cost measure. We aim at constructing a library of anisotropic bases that are suitable for the class of 2-D binarized character images. We consider two, a dyadic and a non-dyadic generalization scheme of the Haar wavelet packets that lead to anisotropic bases. For the non-dyadic case, generalized Fibonacci p-trees are used to derive the space division structure of the transform. Both schemes allow for an efficient O(N log N) best basis search algorithm. The so built extended library of anisotropic Haar bases is used in the problem of optical character recognition. A special case, namely recognition of characters from very low resolution, noisy TV images is investigated. The best Haar basis found is then used in the feature extraction stage of a standard OCR system. We achieve very promising recognition rates for experimental databases of synthetic and real images separated into 59 classes.
Links toward previous steps (curation, corpus...)
- to stream PascalFrancis, to step Corpus: Pour aller vers cette notice dans l'étape Curation :000527
- to stream PascalFrancis, to step Curation: Pour aller vers cette notice dans l'étape Curation :000263
- to stream PascalFrancis, to step Checkpoint: Pour aller vers cette notice dans l'étape Curation :000484
- to stream Main, to step Merge: Pour aller vers cette notice dans l'étape Curation :001695
Links to Exploration step
Pascal:04-0486698Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" level="a">Feature extraction by best anisotropic Haar bases in an OCR system</title>
<author><name sortKey="Gotchev, Atanas" sort="Gotchev, Atanas" uniqKey="Gotchev A" first="Atanas" last="Gotchev">Atanas Gotchev</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Institute of Signal Processing, Tampere University of Technology, P. O. Box 553</s1>
<s2>33101 Tampere</s2>
<s3>FIN</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>Finlande</country>
<wicri:noRegion>33101 Tampere</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Rusanovskyy, Dmytro" sort="Rusanovskyy, Dmytro" uniqKey="Rusanovskyy D" first="Dmytro" last="Rusanovskyy">Dmytro Rusanovskyy</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Institute of Signal Processing, Tampere University of Technology, P. O. Box 553</s1>
<s2>33101 Tampere</s2>
<s3>FIN</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>Finlande</country>
<wicri:noRegion>33101 Tampere</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Popov, Roumen" sort="Popov, Roumen" uniqKey="Popov R" first="Roumen" last="Popov">Roumen Popov</name>
<affiliation wicri:level="1"><inist:fA14 i1="02"><s1>Nokia Research Center, Nokia Group, Summit Avenue</s1>
<s2>Farnborough, Hampshire</s2>
<s3>GBR</s3>
<sZ>3 aut.</sZ>
</inist:fA14>
<country>Royaume-Uni</country>
<wicri:noRegion>Farnborough, Hampshire</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Egiazarian, Karen" sort="Egiazarian, Karen" uniqKey="Egiazarian K" first="Karen" last="Egiazarian">Karen Egiazarian</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Institute of Signal Processing, Tampere University of Technology, P. O. Box 553</s1>
<s2>33101 Tampere</s2>
<s3>FIN</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>Finlande</country>
<wicri:noRegion>33101 Tampere</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Astola, Jaakko" sort="Astola, Jaakko" uniqKey="Astola J" first="Jaakko" last="Astola">Jaakko Astola</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Institute of Signal Processing, Tampere University of Technology, P. O. Box 553</s1>
<s2>33101 Tampere</s2>
<s3>FIN</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>Finlande</country>
<wicri:noRegion>33101 Tampere</wicri:noRegion>
</affiliation>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">INIST</idno>
<idno type="inist">04-0486698</idno>
<date when="2004">2004</date>
<idno type="stanalyst">PASCAL 04-0486698 INIST</idno>
<idno type="RBID">Pascal:04-0486698</idno>
<idno type="wicri:Area/PascalFrancis/Corpus">000527</idno>
<idno type="wicri:Area/PascalFrancis/Curation">000263</idno>
<idno type="wicri:Area/PascalFrancis/Checkpoint">000484</idno>
<idno type="wicri:doubleKey">1017-2653:2004:Gotchev A:feature:extraction:by</idno>
<idno type="wicri:Area/Main/Merge">001695</idno>
<idno type="wicri:Area/Main/Curation">001633</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en" level="a">Feature extraction by best anisotropic Haar bases in an OCR system</title>
<author><name sortKey="Gotchev, Atanas" sort="Gotchev, Atanas" uniqKey="Gotchev A" first="Atanas" last="Gotchev">Atanas Gotchev</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Institute of Signal Processing, Tampere University of Technology, P. O. Box 553</s1>
<s2>33101 Tampere</s2>
<s3>FIN</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>Finlande</country>
<wicri:noRegion>33101 Tampere</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Rusanovskyy, Dmytro" sort="Rusanovskyy, Dmytro" uniqKey="Rusanovskyy D" first="Dmytro" last="Rusanovskyy">Dmytro Rusanovskyy</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Institute of Signal Processing, Tampere University of Technology, P. O. Box 553</s1>
<s2>33101 Tampere</s2>
<s3>FIN</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>Finlande</country>
<wicri:noRegion>33101 Tampere</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Popov, Roumen" sort="Popov, Roumen" uniqKey="Popov R" first="Roumen" last="Popov">Roumen Popov</name>
<affiliation wicri:level="1"><inist:fA14 i1="02"><s1>Nokia Research Center, Nokia Group, Summit Avenue</s1>
<s2>Farnborough, Hampshire</s2>
<s3>GBR</s3>
<sZ>3 aut.</sZ>
</inist:fA14>
<country>Royaume-Uni</country>
<wicri:noRegion>Farnborough, Hampshire</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Egiazarian, Karen" sort="Egiazarian, Karen" uniqKey="Egiazarian K" first="Karen" last="Egiazarian">Karen Egiazarian</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Institute of Signal Processing, Tampere University of Technology, P. O. Box 553</s1>
<s2>33101 Tampere</s2>
<s3>FIN</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>Finlande</country>
<wicri:noRegion>33101 Tampere</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Astola, Jaakko" sort="Astola, Jaakko" uniqKey="Astola J" first="Jaakko" last="Astola">Jaakko Astola</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Institute of Signal Processing, Tampere University of Technology, P. O. Box 553</s1>
<s2>33101 Tampere</s2>
<s3>FIN</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>Finlande</country>
<wicri:noRegion>33101 Tampere</wicri:noRegion>
</affiliation>
</author>
</analytic>
<series><title level="j" type="main">SPIE proceedings series</title>
<idno type="ISSN">1017-2653</idno>
<imprint><date when="2004">2004</date>
</imprint>
</series>
</biblStruct>
</sourceDesc>
<seriesStmt><title level="j" type="main">SPIE proceedings series</title>
<idno type="ISSN">1017-2653</idno>
</seriesStmt>
</fileDesc>
<profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Character recognition</term>
<term>Database</term>
<term>Feature extraction</term>
<term>Haar function</term>
<term>Image quality</term>
<term>Image resolution</term>
<term>Low resolution</term>
<term>Noisy image</term>
<term>Optical character recognition</term>
<term>Pattern recognition</term>
<term>Search algorithm</term>
<term>Signal processing</term>
<term>Television</term>
</keywords>
<keywords scheme="Pascal" xml:lang="fr"><term>Extraction caractéristique</term>
<term>Fonction Haar</term>
<term>Reconnaissance optique caractère</term>
<term>Reconnaissance caractère</term>
<term>Algorithme recherche</term>
<term>Basse résolution</term>
<term>Résolution image</term>
<term>Image bruitée</term>
<term>Télévision</term>
<term>Base donnée</term>
<term>Traitement signal</term>
<term>Reconnaissance forme</term>
<term>Qualité image</term>
</keywords>
<keywords scheme="Wicri" type="topic" xml:lang="fr"><term>Télévision</term>
<term>Base de données</term>
</keywords>
</textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">In this contribution, we explore the best basis paradigm for in feature extraction. According to this paradigm, a library of bases is built and the best basis is found for a given signal class with respect to some cost measure. We aim at constructing a library of anisotropic bases that are suitable for the class of 2-D binarized character images. We consider two, a dyadic and a non-dyadic generalization scheme of the Haar wavelet packets that lead to anisotropic bases. For the non-dyadic case, generalized Fibonacci p-trees are used to derive the space division structure of the transform. Both schemes allow for an efficient O(N log N) best basis search algorithm. The so built extended library of anisotropic Haar bases is used in the problem of optical character recognition. A special case, namely recognition of characters from very low resolution, noisy TV images is investigated. The best Haar basis found is then used in the feature extraction stage of a standard OCR system. We achieve very promising recognition rates for experimental databases of synthetic and real images separated into 59 classes.</div>
</front>
</TEI>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Ticri/CIDE/explor/OcrV1/Data/Main/Curation
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 001633 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Curation/biblio.hfd -nk 001633 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Ticri/CIDE |area= OcrV1 |flux= Main |étape= Curation |type= RBID |clé= Pascal:04-0486698 |texte= Feature extraction by best anisotropic Haar bases in an OCR system }}
![]() | This area was generated with Dilib version V0.6.32. | ![]() |