Consensus Network Based Hypotheses Combination for Arabic Offline Handwriting Recognition

Offline handwriting recognition (OHR) is an extremely challenging task because of many factors including variations in writing style, writing device and material, and noise in the scanning and collection process. Due to the diverse nature of the above challenges, it is highly unlikely that a single...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Prasad, Rohit, Kamali, Matin, Belanger, David, Rosti, Antti-Veikko, Matsoukas, Spyros, Natarajan, Prem
Format: Tagungsbericht
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 2864
container_issue
container_start_page 2861
container_title
container_volume
creator Prasad, Rohit
Kamali, Matin
Belanger, David
Rosti, Antti-Veikko
Matsoukas, Spyros
Natarajan, Prem
description Offline handwriting recognition (OHR) is an extremely challenging task because of many factors including variations in writing style, writing device and material, and noise in the scanning and collection process. Due to the diverse nature of the above challenges, it is highly unlikely that a single recognition technique can address all the characteristics of real-world handwritten documents. Therefore, one must consider designing different systems, each addressing specific challenges in the handwritten corpus, and then combining the hypotheses from these diverse systems. To that end, we present an innovative approach for combining hypotheses from multiple handwriting recognition systems. Our approach is based on generating a consensus network using hypotheses from a diverse set of handwriting recognition systems. Next, we decode the consensus network for producing the best possible hypothesis given an error criterion. Experimental results on an Arabic OHR task show that our combination algorithm outperforms the NIST ROVER technique and results in a 7% relative reduction in the word error rate over the single best OHR system.
doi_str_mv 10.1109/ICPR.2010.701
format Conference Proceeding
fullrecord <record><control><sourceid>ieee_6IE</sourceid><recordid>TN_cdi_ieee_primary_5597041</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>5597041</ieee_id><sourcerecordid>5597041</sourcerecordid><originalsourceid>FETCH-LOGICAL-i90t-44c5f8c251e9652bede97e476294f7b298e660233787b1431dad698b05d261213</originalsourceid><addsrcrecordid>eNo1jMtOwzAUBc1LIpQsWbHxD6T4On7EyxIBrVRRVHXDqrKTm2LR2lUcVPXvKQJWo9EcHULugI0BmHmY1W_LMWcn1QzOSG50BYILoaUAcU4yXpVQ6JNekJv_wPklyYBJKISScE3ylLxjXGmlpZQZea9jSBjSV6KvOBxi_0kfbcKWTo_7OHxgwkTruHM-2MHHQLvY00lvnW_oouu2PiCd2tAeej_4sKFLbOIm-J_pLbnq7DZh_scRWT0_reppMV-8zOrJvPCGDYUQjeyqhktAoyR32KLRKLTiRnTacVOhUoyXpa60A1FCa1tlKsdkyxVwKEfk_vfWI-J63_ud7Y9rKY1mAspvOeJVxg</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>Consensus Network Based Hypotheses Combination for Arabic Offline Handwriting Recognition</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Prasad, Rohit ; Kamali, Matin ; Belanger, David ; Rosti, Antti-Veikko ; Matsoukas, Spyros ; Natarajan, Prem</creator><creatorcontrib>Prasad, Rohit ; Kamali, Matin ; Belanger, David ; Rosti, Antti-Veikko ; Matsoukas, Spyros ; Natarajan, Prem</creatorcontrib><description>Offline handwriting recognition (OHR) is an extremely challenging task because of many factors including variations in writing style, writing device and material, and noise in the scanning and collection process. Due to the diverse nature of the above challenges, it is highly unlikely that a single recognition technique can address all the characteristics of real-world handwritten documents. Therefore, one must consider designing different systems, each addressing specific challenges in the handwritten corpus, and then combining the hypotheses from these diverse systems. To that end, we present an innovative approach for combining hypotheses from multiple handwriting recognition systems. Our approach is based on generating a consensus network using hypotheses from a diverse set of handwriting recognition systems. Next, we decode the consensus network for producing the best possible hypothesis given an error criterion. Experimental results on an Arabic OHR task show that our combination algorithm outperforms the NIST ROVER technique and results in a 7% relative reduction in the word error rate over the single best OHR system.</description><identifier>ISSN: 1051-4651</identifier><identifier>ISBN: 1424475422</identifier><identifier>ISBN: 9781424475421</identifier><identifier>EISSN: 2831-7475</identifier><identifier>EISBN: 9781424475414</identifier><identifier>EISBN: 9780769541099</identifier><identifier>EISBN: 1424475414</identifier><identifier>EISBN: 0769541097</identifier><identifier>DOI: 10.1109/ICPR.2010.701</identifier><language>eng</language><publisher>IEEE</publisher><subject>consensus network ; Decoding ; Error analysis ; handwriting ; Handwriting recognition ; Hidden Markov models ; Lattices ; NIST ; OCR ; ROVER ; Speech recognition ; system combination</subject><ispartof>2010 20th International Conference on Pattern Recognition, 2010, p.2861-2864</ispartof><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/5597041$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,780,784,789,790,2058,27925,54920</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/5597041$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Prasad, Rohit</creatorcontrib><creatorcontrib>Kamali, Matin</creatorcontrib><creatorcontrib>Belanger, David</creatorcontrib><creatorcontrib>Rosti, Antti-Veikko</creatorcontrib><creatorcontrib>Matsoukas, Spyros</creatorcontrib><creatorcontrib>Natarajan, Prem</creatorcontrib><title>Consensus Network Based Hypotheses Combination for Arabic Offline Handwriting Recognition</title><title>2010 20th International Conference on Pattern Recognition</title><addtitle>ICPR</addtitle><description>Offline handwriting recognition (OHR) is an extremely challenging task because of many factors including variations in writing style, writing device and material, and noise in the scanning and collection process. Due to the diverse nature of the above challenges, it is highly unlikely that a single recognition technique can address all the characteristics of real-world handwritten documents. Therefore, one must consider designing different systems, each addressing specific challenges in the handwritten corpus, and then combining the hypotheses from these diverse systems. To that end, we present an innovative approach for combining hypotheses from multiple handwriting recognition systems. Our approach is based on generating a consensus network using hypotheses from a diverse set of handwriting recognition systems. Next, we decode the consensus network for producing the best possible hypothesis given an error criterion. Experimental results on an Arabic OHR task show that our combination algorithm outperforms the NIST ROVER technique and results in a 7% relative reduction in the word error rate over the single best OHR system.</description><subject>consensus network</subject><subject>Decoding</subject><subject>Error analysis</subject><subject>handwriting</subject><subject>Handwriting recognition</subject><subject>Hidden Markov models</subject><subject>Lattices</subject><subject>NIST</subject><subject>OCR</subject><subject>ROVER</subject><subject>Speech recognition</subject><subject>system combination</subject><issn>1051-4651</issn><issn>2831-7475</issn><isbn>1424475422</isbn><isbn>9781424475421</isbn><isbn>9781424475414</isbn><isbn>9780769541099</isbn><isbn>1424475414</isbn><isbn>0769541097</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2010</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><sourceid>RIE</sourceid><recordid>eNo1jMtOwzAUBc1LIpQsWbHxD6T4On7EyxIBrVRRVHXDqrKTm2LR2lUcVPXvKQJWo9EcHULugI0BmHmY1W_LMWcn1QzOSG50BYILoaUAcU4yXpVQ6JNekJv_wPklyYBJKISScE3ylLxjXGmlpZQZea9jSBjSV6KvOBxi_0kfbcKWTo_7OHxgwkTruHM-2MHHQLvY00lvnW_oouu2PiCd2tAeej_4sKFLbOIm-J_pLbnq7DZh_scRWT0_reppMV-8zOrJvPCGDYUQjeyqhktAoyR32KLRKLTiRnTacVOhUoyXpa60A1FCa1tlKsdkyxVwKEfk_vfWI-J63_ud7Y9rKY1mAspvOeJVxg</recordid><startdate>201008</startdate><enddate>201008</enddate><creator>Prasad, Rohit</creator><creator>Kamali, Matin</creator><creator>Belanger, David</creator><creator>Rosti, Antti-Veikko</creator><creator>Matsoukas, Spyros</creator><creator>Natarajan, Prem</creator><general>IEEE</general><scope>6IE</scope><scope>6IL</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIL</scope></search><sort><creationdate>201008</creationdate><title>Consensus Network Based Hypotheses Combination for Arabic Offline Handwriting Recognition</title><author>Prasad, Rohit ; Kamali, Matin ; Belanger, David ; Rosti, Antti-Veikko ; Matsoukas, Spyros ; Natarajan, Prem</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-i90t-44c5f8c251e9652bede97e476294f7b298e660233787b1431dad698b05d261213</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2010</creationdate><topic>consensus network</topic><topic>Decoding</topic><topic>Error analysis</topic><topic>handwriting</topic><topic>Handwriting recognition</topic><topic>Hidden Markov models</topic><topic>Lattices</topic><topic>NIST</topic><topic>OCR</topic><topic>ROVER</topic><topic>Speech recognition</topic><topic>system combination</topic><toplevel>online_resources</toplevel><creatorcontrib>Prasad, Rohit</creatorcontrib><creatorcontrib>Kamali, Matin</creatorcontrib><creatorcontrib>Belanger, David</creatorcontrib><creatorcontrib>Rosti, Antti-Veikko</creatorcontrib><creatorcontrib>Matsoukas, Spyros</creatorcontrib><creatorcontrib>Natarajan, Prem</creatorcontrib><collection>IEEE Electronic Library (IEL) Conference Proceedings</collection><collection>IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume</collection><collection>IEEE Xplore All Conference Proceedings</collection><collection>IEEE Electronic Library (IEL)</collection><collection>IEEE Proceedings Order Plans (POP All) 1998-Present</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Prasad, Rohit</au><au>Kamali, Matin</au><au>Belanger, David</au><au>Rosti, Antti-Veikko</au><au>Matsoukas, Spyros</au><au>Natarajan, Prem</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Consensus Network Based Hypotheses Combination for Arabic Offline Handwriting Recognition</atitle><btitle>2010 20th International Conference on Pattern Recognition</btitle><stitle>ICPR</stitle><date>2010-08</date><risdate>2010</risdate><spage>2861</spage><epage>2864</epage><pages>2861-2864</pages><issn>1051-4651</issn><eissn>2831-7475</eissn><isbn>1424475422</isbn><isbn>9781424475421</isbn><eisbn>9781424475414</eisbn><eisbn>9780769541099</eisbn><eisbn>1424475414</eisbn><eisbn>0769541097</eisbn><abstract>Offline handwriting recognition (OHR) is an extremely challenging task because of many factors including variations in writing style, writing device and material, and noise in the scanning and collection process. Due to the diverse nature of the above challenges, it is highly unlikely that a single recognition technique can address all the characteristics of real-world handwritten documents. Therefore, one must consider designing different systems, each addressing specific challenges in the handwritten corpus, and then combining the hypotheses from these diverse systems. To that end, we present an innovative approach for combining hypotheses from multiple handwriting recognition systems. Our approach is based on generating a consensus network using hypotheses from a diverse set of handwriting recognition systems. Next, we decode the consensus network for producing the best possible hypothesis given an error criterion. Experimental results on an Arabic OHR task show that our combination algorithm outperforms the NIST ROVER technique and results in a 7% relative reduction in the word error rate over the single best OHR system.</abstract><pub>IEEE</pub><doi>10.1109/ICPR.2010.701</doi><tpages>4</tpages></addata></record>
fulltext fulltext_linktorsrc
identifier ISSN: 1051-4651
ispartof 2010 20th International Conference on Pattern Recognition, 2010, p.2861-2864
issn 1051-4651
2831-7475
language eng
recordid cdi_ieee_primary_5597041
source IEEE Electronic Library (IEL) Conference Proceedings
subjects consensus network
Decoding
Error analysis
handwriting
Handwriting recognition
Hidden Markov models
Lattices
NIST
OCR
ROVER
Speech recognition
system combination
title Consensus Network Based Hypotheses Combination for Arabic Offline Handwriting Recognition
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-21T19%3A12%3A06IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-ieee_6IE&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=Consensus%20Network%20Based%20Hypotheses%20Combination%20for%20Arabic%20Offline%20Handwriting%20Recognition&rft.btitle=2010%2020th%20International%20Conference%20on%20Pattern%20Recognition&rft.au=Prasad,%20Rohit&rft.date=2010-08&rft.spage=2861&rft.epage=2864&rft.pages=2861-2864&rft.issn=1051-4651&rft.eissn=2831-7475&rft.isbn=1424475422&rft.isbn_list=9781424475421&rft_id=info:doi/10.1109/ICPR.2010.701&rft_dat=%3Cieee_6IE%3E5597041%3C/ieee_6IE%3E%3Curl%3E%3C/url%3E&rft.eisbn=9781424475414&rft.eisbn_list=9780769541099&rft.eisbn_list=1424475414&rft.eisbn_list=0769541097&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=5597041&rfr_iscdi=true