Eigenfaces vs. Fisherfaces: recognition using class specific linear projection

We develop a face recognition algorithm which is insensitive to large variation in lighting direction and facial expression. Taking a pattern classification approach, we consider each pixel in an image as a coordinate in a high-dimensional space. We take advantage of the observation that the images...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on pattern analysis and machine intelligence 1997-07, Vol.19 (7), p.711-720
Hauptverfasser: Belhumeur, P.N., Hespanha, J.P., Kriegman, D.J.
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 720
container_issue 7
container_start_page 711
container_title IEEE transactions on pattern analysis and machine intelligence
container_volume 19
creator Belhumeur, P.N.
Hespanha, J.P.
Kriegman, D.J.
description We develop a face recognition algorithm which is insensitive to large variation in lighting direction and facial expression. Taking a pattern classification approach, we consider each pixel in an image as a coordinate in a high-dimensional space. We take advantage of the observation that the images of a particular face, under varying illumination but fixed pose, lie in a 3D linear subspace of the high dimensional image space-if the face is a Lambertian surface without shadowing. However, since faces are not truly Lambertian surfaces and do indeed produce self-shadowing, images will deviate from this linear subspace. Rather than explicitly modeling this deviation, we linearly project the image into a subspace in a manner which discounts those regions of the face with large deviation. Our projection method is based on Fisher's linear discriminant and produces well separated classes in a low-dimensional subspace, even under severe variation in lighting and facial expressions. The eigenface technique, another method based on linearly projecting the image space to a low dimensional subspace, has similar computational requirements. Yet, extensive experimental results demonstrate that the proposed "Fisherface" method has error rates that are lower than those of the eigenface technique for tests on the Harvard and Yale face databases.
doi_str_mv 10.1109/34.598228
format Article
fullrecord <record><control><sourceid>proquest_RIE</sourceid><recordid>TN_cdi_ieee_primary_598228</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>598228</ieee_id><sourcerecordid>28307979</sourcerecordid><originalsourceid>FETCH-LOGICAL-c277t-1f1debcfead48ae7772885c2b295127d1e86f5c1050f0fb9f2591310b5fcfca93</originalsourceid><addsrcrecordid>eNo90EFLAzEQBeAgCtbqwaunnAQPWzPJpkm8SWlVKHrR85JNJzVlu1szreC_t3WLp2GYjzfwGLsGMQIQ7l6VI-2slPaEDcApVyit3CkbCBjLwlppz9kF0UoIKLVQA_Y6TUtsow9I_JtGfJboE_Pf_sAzhm7Zpm3qWr6j1C55aDwRpw2GFFPgTWrRZ77J3QrDgV2ys-gbwqvjHLKP2fR98lzM355eJo_zIkhjtgVEWGAdIvpFaT0aY6S1OshaOg3SLADtOOoAQosoYu2i1A4UiFrHEIN3ashu-9z9668d0rZaJwrYNL7FbkeVtEoYZw7wrochd0QZY7XJae3zTwWiOjRWqbLqG9vbm94mRPx3x-Mv9GBnBg</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>28307979</pqid></control><display><type>article</type><title>Eigenfaces vs. Fisherfaces: recognition using class specific linear projection</title><source>IEEE Electronic Library (IEL)</source><creator>Belhumeur, P.N. ; Hespanha, J.P. ; Kriegman, D.J.</creator><creatorcontrib>Belhumeur, P.N. ; Hespanha, J.P. ; Kriegman, D.J.</creatorcontrib><description>We develop a face recognition algorithm which is insensitive to large variation in lighting direction and facial expression. Taking a pattern classification approach, we consider each pixel in an image as a coordinate in a high-dimensional space. We take advantage of the observation that the images of a particular face, under varying illumination but fixed pose, lie in a 3D linear subspace of the high dimensional image space-if the face is a Lambertian surface without shadowing. However, since faces are not truly Lambertian surfaces and do indeed produce self-shadowing, images will deviate from this linear subspace. Rather than explicitly modeling this deviation, we linearly project the image into a subspace in a manner which discounts those regions of the face with large deviation. Our projection method is based on Fisher's linear discriminant and produces well separated classes in a low-dimensional subspace, even under severe variation in lighting and facial expressions. The eigenface technique, another method based on linearly projecting the image space to a low dimensional subspace, has similar computational requirements. Yet, extensive experimental results demonstrate that the proposed "Fisherface" method has error rates that are lower than those of the eigenface technique for tests on the Harvard and Yale face databases.</description><identifier>ISSN: 0162-8828</identifier><identifier>EISSN: 1939-3539</identifier><identifier>DOI: 10.1109/34.598228</identifier><identifier>CODEN: ITPIDJ</identifier><language>eng</language><publisher>IEEE</publisher><subject>Error analysis ; Face detection ; Face recognition ; Light scattering ; Light sources ; Lighting ; Pattern classification ; Pixel ; Principal component analysis ; Shadow mapping</subject><ispartof>IEEE transactions on pattern analysis and machine intelligence, 1997-07, Vol.19 (7), p.711-720</ispartof><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c277t-1f1debcfead48ae7772885c2b295127d1e86f5c1050f0fb9f2591310b5fcfca93</citedby><cites>FETCH-LOGICAL-c277t-1f1debcfead48ae7772885c2b295127d1e86f5c1050f0fb9f2591310b5fcfca93</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/598228$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>314,780,784,796,27924,27925,54758</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/598228$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Belhumeur, P.N.</creatorcontrib><creatorcontrib>Hespanha, J.P.</creatorcontrib><creatorcontrib>Kriegman, D.J.</creatorcontrib><title>Eigenfaces vs. Fisherfaces: recognition using class specific linear projection</title><title>IEEE transactions on pattern analysis and machine intelligence</title><addtitle>TPAMI</addtitle><description>We develop a face recognition algorithm which is insensitive to large variation in lighting direction and facial expression. Taking a pattern classification approach, we consider each pixel in an image as a coordinate in a high-dimensional space. We take advantage of the observation that the images of a particular face, under varying illumination but fixed pose, lie in a 3D linear subspace of the high dimensional image space-if the face is a Lambertian surface without shadowing. However, since faces are not truly Lambertian surfaces and do indeed produce self-shadowing, images will deviate from this linear subspace. Rather than explicitly modeling this deviation, we linearly project the image into a subspace in a manner which discounts those regions of the face with large deviation. Our projection method is based on Fisher's linear discriminant and produces well separated classes in a low-dimensional subspace, even under severe variation in lighting and facial expressions. The eigenface technique, another method based on linearly projecting the image space to a low dimensional subspace, has similar computational requirements. Yet, extensive experimental results demonstrate that the proposed "Fisherface" method has error rates that are lower than those of the eigenface technique for tests on the Harvard and Yale face databases.</description><subject>Error analysis</subject><subject>Face detection</subject><subject>Face recognition</subject><subject>Light scattering</subject><subject>Light sources</subject><subject>Lighting</subject><subject>Pattern classification</subject><subject>Pixel</subject><subject>Principal component analysis</subject><subject>Shadow mapping</subject><issn>0162-8828</issn><issn>1939-3539</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>1997</creationdate><recordtype>article</recordtype><recordid>eNo90EFLAzEQBeAgCtbqwaunnAQPWzPJpkm8SWlVKHrR85JNJzVlu1szreC_t3WLp2GYjzfwGLsGMQIQ7l6VI-2slPaEDcApVyit3CkbCBjLwlppz9kF0UoIKLVQA_Y6TUtsow9I_JtGfJboE_Pf_sAzhm7Zpm3qWr6j1C55aDwRpw2GFFPgTWrRZ77J3QrDgV2ys-gbwqvjHLKP2fR98lzM355eJo_zIkhjtgVEWGAdIvpFaT0aY6S1OshaOg3SLADtOOoAQosoYu2i1A4UiFrHEIN3ashu-9z9668d0rZaJwrYNL7FbkeVtEoYZw7wrochd0QZY7XJae3zTwWiOjRWqbLqG9vbm94mRPx3x-Mv9GBnBg</recordid><startdate>19970701</startdate><enddate>19970701</enddate><creator>Belhumeur, P.N.</creator><creator>Hespanha, J.P.</creator><creator>Kriegman, D.J.</creator><general>IEEE</general><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope></search><sort><creationdate>19970701</creationdate><title>Eigenfaces vs. Fisherfaces: recognition using class specific linear projection</title><author>Belhumeur, P.N. ; Hespanha, J.P. ; Kriegman, D.J.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c277t-1f1debcfead48ae7772885c2b295127d1e86f5c1050f0fb9f2591310b5fcfca93</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>1997</creationdate><topic>Error analysis</topic><topic>Face detection</topic><topic>Face recognition</topic><topic>Light scattering</topic><topic>Light sources</topic><topic>Lighting</topic><topic>Pattern classification</topic><topic>Pixel</topic><topic>Principal component analysis</topic><topic>Shadow mapping</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Belhumeur, P.N.</creatorcontrib><creatorcontrib>Hespanha, J.P.</creatorcontrib><creatorcontrib>Kriegman, D.J.</creatorcontrib><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts – Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>IEEE transactions on pattern analysis and machine intelligence</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Belhumeur, P.N.</au><au>Hespanha, J.P.</au><au>Kriegman, D.J.</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Eigenfaces vs. Fisherfaces: recognition using class specific linear projection</atitle><jtitle>IEEE transactions on pattern analysis and machine intelligence</jtitle><stitle>TPAMI</stitle><date>1997-07-01</date><risdate>1997</risdate><volume>19</volume><issue>7</issue><spage>711</spage><epage>720</epage><pages>711-720</pages><issn>0162-8828</issn><eissn>1939-3539</eissn><coden>ITPIDJ</coden><abstract>We develop a face recognition algorithm which is insensitive to large variation in lighting direction and facial expression. Taking a pattern classification approach, we consider each pixel in an image as a coordinate in a high-dimensional space. We take advantage of the observation that the images of a particular face, under varying illumination but fixed pose, lie in a 3D linear subspace of the high dimensional image space-if the face is a Lambertian surface without shadowing. However, since faces are not truly Lambertian surfaces and do indeed produce self-shadowing, images will deviate from this linear subspace. Rather than explicitly modeling this deviation, we linearly project the image into a subspace in a manner which discounts those regions of the face with large deviation. Our projection method is based on Fisher's linear discriminant and produces well separated classes in a low-dimensional subspace, even under severe variation in lighting and facial expressions. The eigenface technique, another method based on linearly projecting the image space to a low dimensional subspace, has similar computational requirements. Yet, extensive experimental results demonstrate that the proposed "Fisherface" method has error rates that are lower than those of the eigenface technique for tests on the Harvard and Yale face databases.</abstract><pub>IEEE</pub><doi>10.1109/34.598228</doi><tpages>10</tpages></addata></record>
fulltext fulltext_linktorsrc
identifier ISSN: 0162-8828
ispartof IEEE transactions on pattern analysis and machine intelligence, 1997-07, Vol.19 (7), p.711-720
issn 0162-8828
1939-3539
language eng
recordid cdi_ieee_primary_598228
source IEEE Electronic Library (IEL)
subjects Error analysis
Face detection
Face recognition
Light scattering
Light sources
Lighting
Pattern classification
Pixel
Principal component analysis
Shadow mapping
title Eigenfaces vs. Fisherfaces: recognition using class specific linear projection
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-24T22%3A43%3A37IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_RIE&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Eigenfaces%20vs.%20Fisherfaces:%20recognition%20using%20class%20specific%20linear%20projection&rft.jtitle=IEEE%20transactions%20on%20pattern%20analysis%20and%20machine%20intelligence&rft.au=Belhumeur,%20P.N.&rft.date=1997-07-01&rft.volume=19&rft.issue=7&rft.spage=711&rft.epage=720&rft.pages=711-720&rft.issn=0162-8828&rft.eissn=1939-3539&rft.coden=ITPIDJ&rft_id=info:doi/10.1109/34.598228&rft_dat=%3Cproquest_RIE%3E28307979%3C/proquest_RIE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=28307979&rft_id=info:pmid/&rft_ieee_id=598228&rfr_iscdi=true