Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas
Autor(a) principal: | |
---|---|
Data de Publicação: | 2023 |
Tipo de documento: | Dissertação |
Idioma: | por |
Título da fonte: | Biblioteca Digital de Teses e Dissertações do LNCC |
Texto Completo: | https://tede.lncc.br/handle/tede/383 |
Resumo: | Despite the efficiency of principal components analysis (Principal Components Analysis - P CA) as an unsupervised method for dimensionality reduction, its algorithm does not incorporate prior information extracted from the specific domain corresponding to the data. One way to address this limitation of P CA is to use spatial weights computed from the data itself, generating weighting maps, which we also call computational maps. In this context, the main objective of this work is to investigate the efficiency of weighting techniques for the design of principal components in gender and facial expression experi- ments, considering classification and reconstruction problems. Specifically, the methodology consists of generating spatial weights to weight the pixels of the input images, translated in relation to the global average of the data. Then, the weighted data is used as input for the traditional P CA algorithm. The method thus obtained is called weighted P CA. Such a data-driven approach has been employed in the literature through the calculation of linear classifiers and discriminant analysis techniques. In this dissertation, we propose other alternatives and analyze the results obtained in the problems of interest. We consider the following techniques for calculating spatial weights: (a) Shannon entropy computed pixel-by-pixel (H); (b) Inverse of Shannon Entropy (̂H); (c) Jensen-Shannon divergence (JS); (d) Student’s t-test (T ); (e) Hyperplanes computed using Smooth Support Vector Machine (W ). The technique (e) is known in the literature for the method in focus, being chosen as a reference for comparison with our proposals. The maps obtained by method (d) are used in face analysis, however, in different methodologies from that addressed in this dissertation. The application of Student’s t-test, Shannon entropy and its inverse, and Jensen-Shannon divergence to calculate spatial weights to compute the weighted P CA for face recognition is the main contribution of this work. The evaluation of the efficiency of the different principal components obtained from each technique is done through visualization, analysis of image reconstruction results and classification. For the latter, the following classifiers were used: Mahalanobis Distance (DM ) and K-Nearest Neighbors (KN N ), with K = 1. The computed weights may form a noisy weight map, which can introduce artifacts in the reconstruction and interfere with the classification. In order to reduce such effects, the computational maps were processed using a methodology based on quadtree, generating new maps, Hqt,̂ Hqt, JSqt, Tqt e Wqt, to reduce small local variations. The use of quadtree in the processing of spatial maps to compute the weighted P CA is another contribution of this work. The computational experiments were carried out using the FEI face image database, which consists of 400 images, 200 of men and 200 of women, where each subject is pictured with facial expressions of smiling and neutral. Another database used was the FERET one, consisting of 400 images, 214 of men and 186 of women, where each subject is pictured with facial expressions smiling and neutral. The results demonstrated the efficiency of the weighted versions of P CA especially when we applied the KN N classifier. In this case, the T ,̂ H, Hqt and̂ Hqt methods achieve the best results. In the case of reconstruction, there is a predominance of superior results for P CA in relation to the others, followed by the JS and W methods. The experiments were reworked with the statistical maps obtained after processing via quadtree, showing improved classification results for thê H and H methods, and showing improved reconstruction results for the W method. |
id |
LNCC_56b76870519d6ad1df9bee0c221abbd1 |
---|---|
oai_identifier_str |
oai:tede-server.lncc.br:tede/383 |
network_acronym_str |
LNCC |
network_name_str |
Biblioteca Digital de Teses e Dissertações do LNCC |
repository_id_str |
|
spelling |
Giraldi, Gilson AntonioHaddad, Diego BarretoGiraldi, Gilson AntonioOliveira, Jauvane Cavalcante deThomaz, Carlos Eduardohttp://lattes.cnpq.br/7371522980263618Miranda, Laura Costa Pereira2024-02-02T14:58:18Z2023-11-30MIRANDA, L. C. P. Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas. 2023. 254 f. Dissertação (Programa de Pós-Graduação em Modelagem Computacional) - Laboratório Nacional de Computação Científica, Petrópolis, 2023.https://tede.lncc.br/handle/tede/383Despite the efficiency of principal components analysis (Principal Components Analysis - P CA) as an unsupervised method for dimensionality reduction, its algorithm does not incorporate prior information extracted from the specific domain corresponding to the data. One way to address this limitation of P CA is to use spatial weights computed from the data itself, generating weighting maps, which we also call computational maps. In this context, the main objective of this work is to investigate the efficiency of weighting techniques for the design of principal components in gender and facial expression experi- ments, considering classification and reconstruction problems. Specifically, the methodology consists of generating spatial weights to weight the pixels of the input images, translated in relation to the global average of the data. Then, the weighted data is used as input for the traditional P CA algorithm. The method thus obtained is called weighted P CA. Such a data-driven approach has been employed in the literature through the calculation of linear classifiers and discriminant analysis techniques. In this dissertation, we propose other alternatives and analyze the results obtained in the problems of interest. We consider the following techniques for calculating spatial weights: (a) Shannon entropy computed pixel-by-pixel (H); (b) Inverse of Shannon Entropy (̂H); (c) Jensen-Shannon divergence (JS); (d) Student’s t-test (T ); (e) Hyperplanes computed using Smooth Support Vector Machine (W ). The technique (e) is known in the literature for the method in focus, being chosen as a reference for comparison with our proposals. The maps obtained by method (d) are used in face analysis, however, in different methodologies from that addressed in this dissertation. The application of Student’s t-test, Shannon entropy and its inverse, and Jensen-Shannon divergence to calculate spatial weights to compute the weighted P CA for face recognition is the main contribution of this work. The evaluation of the efficiency of the different principal components obtained from each technique is done through visualization, analysis of image reconstruction results and classification. For the latter, the following classifiers were used: Mahalanobis Distance (DM ) and K-Nearest Neighbors (KN N ), with K = 1. The computed weights may form a noisy weight map, which can introduce artifacts in the reconstruction and interfere with the classification. In order to reduce such effects, the computational maps were processed using a methodology based on quadtree, generating new maps, Hqt,̂ Hqt, JSqt, Tqt e Wqt, to reduce small local variations. The use of quadtree in the processing of spatial maps to compute the weighted P CA is another contribution of this work. The computational experiments were carried out using the FEI face image database, which consists of 400 images, 200 of men and 200 of women, where each subject is pictured with facial expressions of smiling and neutral. Another database used was the FERET one, consisting of 400 images, 214 of men and 186 of women, where each subject is pictured with facial expressions smiling and neutral. The results demonstrated the efficiency of the weighted versions of P CA especially when we applied the KN N classifier. In this case, the T ,̂ H, Hqt and̂ Hqt methods achieve the best results. In the case of reconstruction, there is a predominance of superior results for P CA in relation to the others, followed by the JS and W methods. The experiments were reworked with the statistical maps obtained after processing via quadtree, showing improved classification results for thê H and H methods, and showing improved reconstruction results for the W method.Apesar da eficiência da análise de componentes principais (Principal Components Analysis - P CA) como um método não-supervisionado, para redução de dimensionalidade, seu algoritmo não incorpora informações prévias extraídas do domínio específico correspon- dente aos dados. Uma forma de tratar essa limitação do P CA é utilizando pesos espaciais computados através dos próprios dados, gerando mapas de ponderação, os quais tam- bém denominamos mapas computacionais. Neste contexto, o principal objetivo deste trabalho é investigar a eficiência de técnicas de ponderação dos dados para o cálculo de componentes principais em experimentos de gênero e expressão facial, considerando problemas de classificação e reconstrução. Especificamente, a metodologia consiste em gerar pesos espaciais para ponderar os pixels das imagens de entrada do P CA, transla- dadas em relação à média global dos dados. Em seguida, os dados assim ponderados são utilizados como entrada para o algoritmo tradicional do P CA. O método assim obtido é denominado P CA ponderado, aproveitando a nomenclatura da literatura onde essa técnica é denominada Weighted P CA. Tal abordagem baseada em dados foi empregada na literatura via cálculo de classificadores lineares e técnicas de análise discriminante. Nesta dissertação propomos outras alternativas e comparamos os resultados obtidos nos proble- mas de interesse. Estaremos considerando as seguintes técnicas para o cálculo dos pesos espaciais: (a) Entropia de Shannon calculada pixel-a-pixel (H); (b) Inverso da Entropia de Shannon (̂H); (c) Divergência Jensen-Shannon (JS); (d) Teste t de Student (T ); (e) Hiperplanos computados usando Smooth Support Vector Machine (W ). A técnica (e) é conhecida na literatura para o método em foco, sendo escolhida como uma referência para comparação com as demais. Os mapas obtidos pelo método (d) são utilizados em análise de faces, porém, em metodologias distintas da metodologia abordada nesta dissertação. A aplicação do Teste t de Student, entropia de Shannon e seu inverso, e divergência de Jensen-Shannon para cálculo de pesos espaciais para cômputo do P CA ponderado em reconhecimento de faces consiste na principal contribuição deste trabalho. A avaliação da eficiência das diferentes componentes principais obtidas com cada técnica é feita através da visualização de componentes principais, análise dos resultados da reconstrução de imagens e classificação. Para esta última, foram utilizados os seguintes classificadores: Distância de Mahalanobis (DM ) e K-Vizinhos Mais Próximos (KN N ), com K = 1. Os pesos computados podem formar um mapa de ponderação ruidoso, o que pode acarretar artefatos na reconstrução e interferir na classificação. Com o objetivo de reduzir tais efeitos os mapas computacionais foram processados utilizando uma metodologia baseada em quadtree, gerando novos mapas Hqt,̂ Hqt, JSqt, Tqt e Wqt, para reduzir pequenas variações locais. A utilização de quadtree no processamento de mapas espaciais para computar o P CA ponderado é outra contribuição deste trabalho. Os experimentos computacionais foram realizados usando a base de imagens de faces da FEI, que consiste de 400 imagens, sendo 200 de homens e 200 de mulheres, em cada caso, com expressões faciais sorrindo e neutra, e a base de imagens FERET, consistindo em 400 imagens, com 214 imagens de homens e 186 imagens de mulheres, em cada caso, com expressões faciais sorrindo e neutra. Os resultados mostraram a eficiência das versões ponderadas do P CA especialmente quando aplicamos o classificador KN N . Neste caso, os métodos T ,̂ H, Hqt ê Hqt foram os melhores. No caso da reconstrução, existe predominância de resultados superiores para o P CA em relação aos demais, seguido dos métodos JS e W . Os experimentos foram refeitos com os mapas obtidos após processamento via quadtree, mostrando melhora dos resultado da classificação para os métodoŝ H e H e mostrando melhora dos resultados da reconstrução para o método W .Submitted by Patrícia Vieira Silva (library@lncc.br) on 2024-02-02T14:57:23Z No. of bitstreams: 1 Dissertacao_Laura_Miranda__final1.pdf: 31517881 bytes, checksum: 8443df0b1f73a7dd7d8b0204adfa443f (MD5)Approved for entry into archive by Patrícia Vieira Silva (library@lncc.br) on 2024-02-02T14:57:59Z (GMT) No. of bitstreams: 1 Dissertacao_Laura_Miranda__final1.pdf: 31517881 bytes, checksum: 8443df0b1f73a7dd7d8b0204adfa443f (MD5)Made available in DSpace on 2024-02-02T14:58:18Z (GMT). No. of bitstreams: 1 Dissertacao_Laura_Miranda__final1.pdf: 31517881 bytes, checksum: 8443df0b1f73a7dd7d8b0204adfa443f (MD5) Previous issue date: 2023-11-30Coordenação de Aperfeiçoamento de Pessoal de Nível Superiorapplication/pdfhttp://tede-server.lncc.br:8080/retrieve/1709/Dissertacao_Laura_Miranda__final1.pdf.jpgporLaboratório Nacional de Computação CientíficaPrograma de Pós-Graduação em Modelagem ComputacionalLNCCBrasilCoordenação de Pós-Graduação e Aperfeiçoamento (COPGA)https://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/openAccessProcessamento de imagens - Modelos matemáticosReconhecimento de facesAnálise de componentes principaisCNPQ::ENGENHARIASAbordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanasinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisreponame:Biblioteca Digital de Teses e Dissertações do LNCCinstname:Laboratório Nacional de Computação Científica (LNCC)instacron:LNCCTHUMBNAILDissertacao_Laura_Miranda__final1.pdf.jpgDissertacao_Laura_Miranda__final1.pdf.jpgimage/jpeg3964http://tede-server.lncc.br:8080/tede/bitstream/tede/383/4/Dissertacao_Laura_Miranda__final1.pdf.jpg08cc9dad1ab5ef61263969da0fe0d7e9MD54TEXTDissertacao_Laura_Miranda__final1.pdf.txtDissertacao_Laura_Miranda__final1.pdf.txttext/plain461088http://tede-server.lncc.br:8080/tede/bitstream/tede/383/3/Dissertacao_Laura_Miranda__final1.pdf.txt8a742765a556dcf414577a65c97ad7c3MD53ORIGINALDissertacao_Laura_Miranda__final1.pdfDissertacao_Laura_Miranda__final1.pdfapplication/pdf31517881http://tede-server.lncc.br:8080/tede/bitstream/tede/383/2/Dissertacao_Laura_Miranda__final1.pdf8443df0b1f73a7dd7d8b0204adfa443fMD52LICENSElicense.txtlicense.txttext/plain; charset=utf-82165http://tede-server.lncc.br:8080/tede/bitstream/tede/383/1/license.txtbd3efa91386c1718a7f26a329fdcb468MD51tede/3832024-02-03 02:37:46.425oai:tede-server.lncc.br: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Biblioteca Digital de Teses e Dissertaçõeshttps://tede.lncc.br/PUBhttps://tede.lncc.br/oai/requestlibrary@lncc.br||library@lncc.bropendoar:2024-02-03T04:37:46Biblioteca Digital de Teses e Dissertações do LNCC - Laboratório Nacional de Computação Científica (LNCC)false |
dc.title.por.fl_str_mv |
Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas |
title |
Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas |
spellingShingle |
Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas Miranda, Laura Costa Pereira Processamento de imagens - Modelos matemáticos Reconhecimento de faces Análise de componentes principais CNPQ::ENGENHARIAS |
title_short |
Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas |
title_full |
Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas |
title_fullStr |
Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas |
title_full_unstemmed |
Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas |
title_sort |
Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas |
author |
Miranda, Laura Costa Pereira |
author_facet |
Miranda, Laura Costa Pereira |
author_role |
author |
dc.contributor.advisor1.fl_str_mv |
Giraldi, Gilson Antonio |
dc.contributor.advisor2.fl_str_mv |
Haddad, Diego Barreto |
dc.contributor.referee1.fl_str_mv |
Giraldi, Gilson Antonio |
dc.contributor.referee2.fl_str_mv |
Oliveira, Jauvane Cavalcante de |
dc.contributor.referee3.fl_str_mv |
Thomaz, Carlos Eduardo |
dc.contributor.authorLattes.fl_str_mv |
http://lattes.cnpq.br/7371522980263618 |
dc.contributor.author.fl_str_mv |
Miranda, Laura Costa Pereira |
contributor_str_mv |
Giraldi, Gilson Antonio Haddad, Diego Barreto Giraldi, Gilson Antonio Oliveira, Jauvane Cavalcante de Thomaz, Carlos Eduardo |
dc.subject.por.fl_str_mv |
Processamento de imagens - Modelos matemáticos Reconhecimento de faces Análise de componentes principais |
topic |
Processamento de imagens - Modelos matemáticos Reconhecimento de faces Análise de componentes principais CNPQ::ENGENHARIAS |
dc.subject.cnpq.fl_str_mv |
CNPQ::ENGENHARIAS |
description |
Despite the efficiency of principal components analysis (Principal Components Analysis - P CA) as an unsupervised method for dimensionality reduction, its algorithm does not incorporate prior information extracted from the specific domain corresponding to the data. One way to address this limitation of P CA is to use spatial weights computed from the data itself, generating weighting maps, which we also call computational maps. In this context, the main objective of this work is to investigate the efficiency of weighting techniques for the design of principal components in gender and facial expression experi- ments, considering classification and reconstruction problems. Specifically, the methodology consists of generating spatial weights to weight the pixels of the input images, translated in relation to the global average of the data. Then, the weighted data is used as input for the traditional P CA algorithm. The method thus obtained is called weighted P CA. Such a data-driven approach has been employed in the literature through the calculation of linear classifiers and discriminant analysis techniques. In this dissertation, we propose other alternatives and analyze the results obtained in the problems of interest. We consider the following techniques for calculating spatial weights: (a) Shannon entropy computed pixel-by-pixel (H); (b) Inverse of Shannon Entropy (̂H); (c) Jensen-Shannon divergence (JS); (d) Student’s t-test (T ); (e) Hyperplanes computed using Smooth Support Vector Machine (W ). The technique (e) is known in the literature for the method in focus, being chosen as a reference for comparison with our proposals. The maps obtained by method (d) are used in face analysis, however, in different methodologies from that addressed in this dissertation. The application of Student’s t-test, Shannon entropy and its inverse, and Jensen-Shannon divergence to calculate spatial weights to compute the weighted P CA for face recognition is the main contribution of this work. The evaluation of the efficiency of the different principal components obtained from each technique is done through visualization, analysis of image reconstruction results and classification. For the latter, the following classifiers were used: Mahalanobis Distance (DM ) and K-Nearest Neighbors (KN N ), with K = 1. The computed weights may form a noisy weight map, which can introduce artifacts in the reconstruction and interfere with the classification. In order to reduce such effects, the computational maps were processed using a methodology based on quadtree, generating new maps, Hqt,̂ Hqt, JSqt, Tqt e Wqt, to reduce small local variations. The use of quadtree in the processing of spatial maps to compute the weighted P CA is another contribution of this work. The computational experiments were carried out using the FEI face image database, which consists of 400 images, 200 of men and 200 of women, where each subject is pictured with facial expressions of smiling and neutral. Another database used was the FERET one, consisting of 400 images, 214 of men and 186 of women, where each subject is pictured with facial expressions smiling and neutral. The results demonstrated the efficiency of the weighted versions of P CA especially when we applied the KN N classifier. In this case, the T ,̂ H, Hqt and̂ Hqt methods achieve the best results. In the case of reconstruction, there is a predominance of superior results for P CA in relation to the others, followed by the JS and W methods. The experiments were reworked with the statistical maps obtained after processing via quadtree, showing improved classification results for thê H and H methods, and showing improved reconstruction results for the W method. |
publishDate |
2023 |
dc.date.issued.fl_str_mv |
2023-11-30 |
dc.date.accessioned.fl_str_mv |
2024-02-02T14:58:18Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/masterThesis |
format |
masterThesis |
status_str |
publishedVersion |
dc.identifier.citation.fl_str_mv |
MIRANDA, L. C. P. Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas. 2023. 254 f. Dissertação (Programa de Pós-Graduação em Modelagem Computacional) - Laboratório Nacional de Computação Científica, Petrópolis, 2023. |
dc.identifier.uri.fl_str_mv |
https://tede.lncc.br/handle/tede/383 |
identifier_str_mv |
MIRANDA, L. C. P. Abordagens computacionais para calcular componentes principais ponderadas com aplicações em análise de imagens de faces humanas. 2023. 254 f. Dissertação (Programa de Pós-Graduação em Modelagem Computacional) - Laboratório Nacional de Computação Científica, Petrópolis, 2023. |
url |
https://tede.lncc.br/handle/tede/383 |
dc.language.iso.fl_str_mv |
por |
language |
por |
dc.rights.driver.fl_str_mv |
https://creativecommons.org/licenses/by-nc-nd/4.0/ info:eu-repo/semantics/openAccess |
rights_invalid_str_mv |
https://creativecommons.org/licenses/by-nc-nd/4.0/ |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
application/pdf |
dc.publisher.none.fl_str_mv |
Laboratório Nacional de Computação Científica |
dc.publisher.program.fl_str_mv |
Programa de Pós-Graduação em Modelagem Computacional |
dc.publisher.initials.fl_str_mv |
LNCC |
dc.publisher.country.fl_str_mv |
Brasil |
dc.publisher.department.fl_str_mv |
Coordenação de Pós-Graduação e Aperfeiçoamento (COPGA) |
publisher.none.fl_str_mv |
Laboratório Nacional de Computação Científica |
dc.source.none.fl_str_mv |
reponame:Biblioteca Digital de Teses e Dissertações do LNCC instname:Laboratório Nacional de Computação Científica (LNCC) instacron:LNCC |
instname_str |
Laboratório Nacional de Computação Científica (LNCC) |
instacron_str |
LNCC |
institution |
LNCC |
reponame_str |
Biblioteca Digital de Teses e Dissertações do LNCC |
collection |
Biblioteca Digital de Teses e Dissertações do LNCC |
bitstream.url.fl_str_mv |
http://tede-server.lncc.br:8080/tede/bitstream/tede/383/4/Dissertacao_Laura_Miranda__final1.pdf.jpg http://tede-server.lncc.br:8080/tede/bitstream/tede/383/3/Dissertacao_Laura_Miranda__final1.pdf.txt http://tede-server.lncc.br:8080/tede/bitstream/tede/383/2/Dissertacao_Laura_Miranda__final1.pdf http://tede-server.lncc.br:8080/tede/bitstream/tede/383/1/license.txt |
bitstream.checksum.fl_str_mv |
08cc9dad1ab5ef61263969da0fe0d7e9 8a742765a556dcf414577a65c97ad7c3 8443df0b1f73a7dd7d8b0204adfa443f bd3efa91386c1718a7f26a329fdcb468 |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 MD5 |
repository.name.fl_str_mv |
Biblioteca Digital de Teses e Dissertações do LNCC - Laboratório Nacional de Computação Científica (LNCC) |
repository.mail.fl_str_mv |
library@lncc.br||library@lncc.br |
_version_ |
1797683220283654144 |