Tesis sobre el tema "Biométrie faciale"

La Reconnaissance automatique des personnes est devenue un sujet de plus en plus important avec l'augmentation constante des besoins en sécurité. De nombreux systèmes biométriques existent. Ils utilisent différentes caractéristiques humaines. Parmi tous les traits biométriques, la reconnaissance faciale inclut des aspects positifs en termes d'accessibilité et de fiabilité. Dans cette thèse, deux défis en reconnaissance faciales sont étudiés. Le premier est le leurrage. Le leurrage en reconnaissance faciale est présenté. Des contre-mesures permettant d'améliorer les systèmes actuels sont proposés. A cet effet, les attaques basées sur des photographies 2D ou des masques 3D sont analysées. Le second défi exploré dans cette thèse est lié aux variations dues à des altérations du visage (i.e. chirurgie plastique), maquillage et accessoires pour le visage (e.g. occultations par la présence de lunettes). L'impact de ces variations en reconnaissance de visage est étudiée séparément. Ensuite, des techniques robustes contre les variations de camouflage sont proposées
Human recognition has become an important topic as the need and investments for security applications grow continuously. Numerous biometric systems exist which utilize various human characteristics. Among all biometrics traits, face recognition is advantageous in terms of accessibility and reliability. In the thesis, two challenges in face recognition are analyzed. The first one is face spoofing. Spoofing in face recognition is explained together with the countermeasure techniques that are proposed for the protection of face recognition systems against spoofing attacks. For this purpose, both 2D photograph and 3D mask attacks are analyzed. The second challenge explored in the thesis is disguise variations, which are due to facial alterations, facial makeup and facial accessories (occlusions). The impact of these disguise variations on face recognition is explored, separately. Then, techniques which are robust against disguise variations are proposed

La Reconnaissance automatique des personnes est devenue un sujet de plus en plus important avec l'augmentation constante des besoins en sécurité. De nombreux systèmes biométriques existent. Ils utilisent différentes caractéristiques humaines. Parmi tous les traits biométriques, la reconnaissance faciale inclut des aspects positifs en termes d'accessibilité et de fiabilité. Dans cette thèse, deux défis en reconnaissance faciales sont étudiés. Le premier est le leurrage. Le leurrage en reconnaissance faciale est présenté. Des contre-mesures permettant d'améliorer les systèmes actuels sont proposés. A cet effet, les attaques basées sur des photographies 2D ou des masques 3D sont analysées. Le second défi exploré dans cette thèse est lié aux variations dues à des altérations du visage (i.e. chirurgie plastique), maquillage et accessoires pour le visage (e.g. occultations par la présence de lunettes). L'impact de ces variations en reconnaissance de visage est étudiée séparément. Ensuite, des techniques robustes contre les variations de camouflage sont proposées
Human recognition has become an important topic as the need and investments for security applications grow continuously. Numerous biometric systems exist which utilize various human characteristics. Among all biometrics traits, face recognition is advantageous in terms of accessibility and reliability. In the thesis, two challenges in face recognition are analyzed. The first one is face spoofing. Spoofing in face recognition is explained together with the countermeasure techniques that are proposed for the protection of face recognition systems against spoofing attacks. For this purpose, both 2D photograph and 3D mask attacks are analyzed. The second challenge explored in the thesis is disguise variations, which are due to facial alterations, facial makeup and facial accessories (occlusions). The impact of these disguise variations on face recognition is explored, separately. Then, techniques which are robust against disguise variations are proposed

La biométrie du visage a suscité, ces derniers temps, l'intérêt grandissant de la communauté scientifique et des industriels de la biométrie vue son caractère naturel, sans contact et non-intrusif. Néanmoins, les performances des systèmes basés sur les images 2D sont affectées par différents types de variabilités comme la pose, les conditions d'éclairage, les occultations et les expressions faciales. Avec la disponibilité de caméras 3D capables d'acquérir la forme tridimensionnelle, moins sensibles aux changements d'illumination et de pose, plusieurs travaux de recherche se sont tournés vers l'étude de cette nouvelle modalité. En revanche, d'autres défis apparaissent comme les déformations de la forme faciales causées par les expressions et le temps de calcul que requièrent les approches développées. Cette thèse s'inscrit dans ce paradigme en proposant de coupler la géométrie Riemannienne avec les techniques d'apprentissage pour une biométrie faciale 3D efficace et robuste aux changements d'expressions. Après une étape de pré-traitement, nous proposons de représenter les surfaces faciales par des collections de courbes 3D qui captent localement leurs formes. Nous utilisons un cadre géométrique existant pour obtenir les déformations " optimales " entre les courbes ainsi que les distances les séparant sur une variété Riemannienne (espace des formes des courbes). Nous appliquons, par la suite, des techniques d'apprentissage afin de déterminer les courbes les plus pertinentes pour deux applications de la biométrie du visage : la reconnaissance d'identité et la classification du genre. Les résultats obtenus sur le benchmark de référence FRGC v2 et leurs comparaison avec les travaux de l'état de l'art confirment tout l'intérêt de coupler l'analyse locale de la forme par une approche géométrique (possibilité de calculer des moyennes, etc.) avec des techniques d'apprentissage (Basting, etc.) pour gagner en temps de calcul et en performances.

La technologie de l'imagerie thermique a largement évolué au cours des deux dernières décennies, grâce aux caméras thermiques qui sont devenues plus abordables et simple à utiliser. Cependant, et étant donné que l'exploration de l'imagerie thermique est relativement nouvelle, seules quelques bases de données publiques sont accessibles à la communauté de recherche. Cette limitation empêche donc l'impact des technologies d'apprentissage profond de générer des systèmes fiables de reconnaissance faciale adaptés au spectre thermique. En essayant de surmonter ces contraintes, les travaux de recherche présentés dans ce manuscrit visent à explorer la synthèse interspectrale pour une intégration efficace et rapide de la technologie thermique dans les systèmes de biométrie faciale existants. Comme première contribution, une nouvelle base de données, contenant des paires d'images de visages visibles et thermiques acquises simultanément, a été collectée et mise en public afin de favoriser la recherche dans le domaine de l’imagerie thermique de visage. Motivé par le besoin d'une intégration simple dans les systèmes de biométrie faciale existants, un ensemble de contributions a proposé un cadre de reconnaissance faciale cross-spectral basé sur une nouvelle approche de synthèse des visages afin d'estimer le visage visible à partir d’un visage thermique. Autres contributions consistant à explorer la synthèse interspectrale, du spectre visible au spectre thermique, pour des tâches de traitement d'images faciales liées à la reconnaissance faciale, sont également présentées notamment la détection des points caractéristiques de visage et l'usurpation d’identité dans le spectre thermique
Thermal imaging technology has significantly evolved during the last couple of decades, mostly thanks to thermal cameras having become more affordable and user friendly. However, and given that the exploration of thermal imagery is reasonably new, only a few public databases are available to the research community. This limitation consequently prevents the impact of deep learning technologies from generating improved and reliable face biometric systems that operate in the thermal spectrum. A possible solution relates to the development of technologies that bridge the gap between visible and thermal spectra. In attempting to respond to this necessity, the research presented in this dissertation aims to explore interspectral synthesis as a direction for efficient and prompt integration of thermal technology in already deployed face biometric systems.As a first contribution, a new database, containing paired visible and thermal face images acquired simultaneously, was collected and made publicly available to foster research in thermal face image processing. Motivated by the need for fast and straightforward integration into existing face recognition systems, a set of contributions consisted in proposing a cross-spectrum face recognition framework based on a novel approach of thermal-to-visible face synthesis in order to estimate the visible face from the thermal input. Contributions consisting in exploring interspectral synthesis from visible to thermal spectrum for facial image processing tasks related to, but different than face recognition, are also presented including facial landmark detection and face biometric spoofing in thermal spectrum

Cette thèse s'intéresse aux biométries dites douces, et notamment à leurs utilisations en termes de sécurité, dans le cadre de différents scénarii commerciaux, incluant des aspects usage. L'accent sera ainsi porté sur les caractéristiques faciales qui constituent un jeu de traits significatifs de l'apparence physique mais aussi comportementale de l'utilisateur permettant de différencier, classer et identifier les individus. Ces traits, qui sont l'âge, le sexe, les cheveux, la peau et la couleur des yeux, mais aussi la présence de lunettes, de moustache ou de barbe, comportent plusieurs avantages notamment la facilité avec laquelle ils peuvent être acquis, mais également du fait qu'ils correspondent à la façon dont les êtres humains perçoivent leurs environnements. Plus précisément, les traits issus de la biométrie douce sont compatibles avec la manière dont l'humain tend à catégoriser son entourage, une démarche impliquant une structuration hiérarchique des différents traits. Cette thèse explore ces différents traits et leurs applications dans les systèmes de biométries douces (SBS), et met l'accent sur la manière dont de tels systèmes peuvent atteindre des buts différents, y compris la recherche accélérée dans des bases de données, l'identification et la ré-identification d'individus, mais également la prédiction et la quantification de l'esthétique d'un visage. Ce travail est motivé notamment par l'importance croissante de ces applications dans notre société en constante évolution, mais aussi par le côté peu contraignant du système. En effet, les SBS sont généralement non-intrusifs, et nécessitent le plus souvent de faibles temps de calculs, permettant ainsi une analyse biométrique rapide, sans imposer obligatoirement l'accord et la coopération de l'individu. Ces atouts rendent la biométrie douce indispensable dans les applications qui ont besoin de traitement d'images ou de vidéos en temps réel.

Cette thèse s’intéresse aux biométries dites douces, et notamment à leurs utilisations en termes de sécurité, dans le cadre de différents scénarii commerciaux, incluant des aspects usage. L'accent sera ainsi porté sur les caractéristiques faciales qui constituent un jeu de traits significatifs de l’apparence physique mais aussi comportementale de l’utilisateur permettant de différencier, classer et identifier les individus. Ces traits, qui sont l'âge, le sexe, les cheveux, la peau et la couleur des yeux, mais aussi la présence de lunettes, de moustache ou de barbe, comportent plusieurs avantages notamment la facilité avec laquelle ils peuvent être acquis, mais également du fait qu’ils correspondent à la façon dont les êtres humains perçoivent leurs environnements. Plus précisément, les traits issus de la biométrie douce sont compatibles avec la manière dont l’humain tend à catégoriser son entourage, une démarche impliquant une structuration hiérarchique des différents traits. Cette thèse explore ces différents traits et leurs applications dans les systèmes de biométries douces (SBS), et met l’accent sur la manière dont de tels systèmes peuvent atteindre des buts différents, y compris la recherche accélérée dans des bases de données, l'identification et la ré-identification d’individus, mais également la prédiction et la quantification de l'esthétique d’un visage. Ce travail est motivé notamment par l'importance croissante de ces applications dans notre société en constante évolution, mais aussi par le côté peu contraignant du système. En effet, les SBS sont généralement nonintrusifs, et nécessitent le plus souvent de faibles temps de calculs
This dissertation studies soft biometrics traits, their applicability in different security and commercial scenarios, as well as related usability aspects. We place the emphasis on human facial soft biometric traits which constitute the set of physical, adhered or behavioral human characteristics that can partially differentiate, classify and identify humans. Such traits, which include characteristics like age, gender, skin and eye color, the presence of glasses, moustache or beard, inherit several advantages such as ease of acquisition, as well as a natural compatibility with how humans perceive their surroundings. Specifically, soft biometric traits are compatible with the human process of classifying and recalling our environment, a process which involves constructions of hierarchical structures of different refined traits. This thesis explores these traits, and their application in soft biometric systems (SBSs), and specifically focuses on how such systems can achieve different goals including database search pruning, human identification, human re–identification and, on a different note, prediction and quantification of facial aesthetics. Our motivation originates from the emerging importance of such applications in our evolving society, as well as from the practicality of such systems. SBSs generally benefit from the non-intrusive nature of acquiring soft biometric traits, and enjoy computational efficiency which in turn allows for fast, enrolment–free and pose–flexible biometric analysis, even in the absence of consent and cooperation by the involved human subjects

Puisque les traits de biométrie douce peuvent fournir des preuves supplémentaires pour aider à déterminer précisément l’identité de l’homme, il y a eu une attention croissante sur la reconnaissance faciale basée sur les biométrie douce ces dernières années. Parmi tous les biométries douces, le sexe et l’ethnicité sont les deux caractéristiques démographiques importantes pour les êtres humains et ils jouent un rôle très fondamental dans l’analyse de visage automatique. En attendant, la reconnaissance des expressions faciales est un autre challenge dans le domaine de l’analyse de visage en raison de la diversité et de l’hybridité des expressions humaines dans différentes cultures, genres et contextes. Ce thèse est dédié à combiner la texture du visage 2D et la morphologie du visage 3D pour estimer les biométries douces: le sexe, l’ethnicité, etc., et reconnaître les expressions faciales. Pour la reconnaissance du sexe et de l’ethnicité, nous présentons une approche efficace en combinant à la fois des textures locales et des caractéristiques de forme extraites à partir des modèles de visage 3D, contrairement aux méthodes existantes qui ne dépendent que des textures ou des caractéristiques de forme. Afin de souligne exhaustivement la différence entre les groupes sexuels et ethniques, nous proposons un nouveau descripteur, à savoir local circular patterns (LCP). Ce descripteur améliore Les motifs binaires locaux (LBP) et ses variantes en remplaçant la quantification binaire par une quantification basée sur le regroupement, entraînant d’une puissance plus discriminative et une meilleure résistance au bruit. En même temps, l’algorithme Adaboost est engagé à sélectionner les caractéristiques discriminatives fortement liés au sexe et à l’ethnicité. Les résultats expérimentaux obtenus sur les bases de données FRGC v2.0 et BU-3DFE démontrent clairement les avantages de la méthode proposée. Pour la reconnaissance des expressions faciales, nous présentons une méthode automatique basée sur les multi-modalité 2D + 3D et démontrons sa performance sur la base des données BU-3DFE. Notre méthode combine des textures locales et des descripteurs de formes pour atteindre l’efficacité et la robustesse. Tout d’abord, un grand ensemble des points des caractéristiques d’images 2D et de modèles 3D sont localisés à l’aide d’un nouvel algorithme, à savoir la cascade parallèle incrémentielle de régression linéaire (iPar-CLR). Ensuite, on utilise un nouveau descripteur basé sur les histogrammes des gradients d’ordre secondaire (HSOG) en conjonction avec le descripteur SIFT pour décrire la texture locale autour de chaque point de caractéristique 2D. De même, la géométrie locale autour de chaque point de caractéristique 3D est décrite par deux nouveaux descripteurs de forme construits à l’aide des quantités différentielle de géométries de la surface au premier ordre et au second ordre, à savoir meshHOG et meshHOS. Enfin, les résultats de reconnaissance des descripteurs 2D et 3D fournis par le classifier SVM sont fusionnés à la fois au niveau de fonctionnalité et de score pour améliorer la précision. Les expérimentaux résultats démontrent clairement qu’il existe des caractéristiques complémentaires entre les descripteurs 2D et 3D. Notre approche basée sur les multi-modalités surpasse les autres méthodes de l’état de l’art en obtenant une précision de reconnaissance 86,32%. De plus, une bonne capacité de généralisation est aussi présentée sur la base de données Bosphorus
Since soft biometrics traits can provide sufficient evidence to precisely determine the identity of human, there has been increasing attention for face based soft biometrics identification in recent years. Among those face based soft biometrics, gender and ethnicity are both key demographic attributes of human beings and they play a very fundamental and important role in automatic machine based face analysis. Meanwhile, facial expression recognition is another challenge problem in face analysis because of the diversity and hybridity of human expressions among different subjects in different cultures, genders and contexts. This Ph.D thesis work is dedicated to combine 2D facial Texture and 3D face morphology for estimating people’s soft biometrics: gender, ethnicity, etc., and recognizing facial expression. For the gender and ethnicity recognition, we present an effective and efficient approach on this issue by combining both boosted local texture and shape features extracted from 3D face models, in contrast to the existing ones that only depend on either 2D texture or 3D shape of faces. In order to comprehensively represent the difference between different genders or ethnics groups, we propose a novel local descriptor, namely local circular patterns (LCP). LCP improves the widely utilized local binary patterns (LBP) and its variants by replacing the binary quantization with a clustering based one, resulting in higher discriminative power as well as better robustness to noise. Meanwhile, the following Adaboost based feature selection finds the most discriminative gender- and ethnic-related features and assigns them with different weights to highlight their importance in classification, which not only further raises the performance but reduces the time and memory cost as well. Experimental results achieved on the FRGC v2.0 and BU-3DFE data sets clearly demonstrate the advantages of the proposed method. For facial expression recognition, we present a fully automatic multi-modal 2D + 3D feature-based facial expression recognition approach and demonstrate its performance on the BU–3DFE database. Our approach combines multi-order gradientbased local texture and shape descriptors in order to achieve efficiency a nd robustness. First, a large set of fiducial facial landmarks of 2D face images along with their 3D face scans are localized using a novel algorithm namely incremental Parallel Cascade of Linear Regression (iPar–CLR). Then, a novel Histogram of Second Order Gradients (HSOG) based local image descriptor in conjunction with the widely used first-order gradient based SIFT descriptor are employed to describe the local texture around each 2D landmark. Similarly, the local geometry around each 3D landmark is described by two novel local shape descriptors constructed using the first-order and the second-order surface differential geometry quantities, i.e., Histogram of mesh Gradients (meshHOG) and Histogram of mesh Shape index (curvature quantization, meshHOS). Finally, the Support Vector Machine (SVM) based recognition results of all 2D and 3D descriptors are fused at both featurelevel and score-level to further improve the accuracy. Comprehensive experimental results demonstrate that there exist impressive complementary characteristics between the 2D and 3D descriptors. We use the BU–3DFE benchmark to compare our approach to the state-of-the-art ones. Our multi-modal feature-based approach outperforms the others by achieving an average recognition accuracy of 86,32%. Moreover, a good generalization ability is shown on the Bosphorus database

El presente informe de experiencia profesional describe el desarrollo de la aplicación “RENIEC MÓVIL FACIAL”, que se implementó debido a que en el año 2016 RENIEC identificó una serie de problemas en sus agencias, entre algunas de ellas, las largas colas que los ciudadanos hacían diariamante para realizar trámites, y otras consultas. Reniec en busca de una solución por la gran demanda de la población, y estando a la vanguardia de las nuevas tecnologías, vio conveniente implementar una aplicación que le permita al ciudadano realizar un trámite y/o consulta de un DNI/DNIe a través de su teléfono móvil, desde cualquier lugar donde se encuentre. Gracias a esta implementación, se logró reducir la gran afluencia de los ciudadanos en las oficinas, brindando un servicio rápido y efectivo.

Ces dernières années, la biométrie a fait l’objet d’une grande attention en raison du besoin sans cesse croissant d’authentification d’identité, notamment pour sécuriser de plus en plus d’applications enlignes. Parmi divers traits biométriques, le visage offre des avantages compétitifs sur les autres, e.g., les empreintes digitales ou l’iris, car il est naturel, non-intrusif et facilement acceptable par les humains. Aujourd’hui, les techniques conventionnelles de reconnaissance faciale ont atteint une performance quasi-parfaite dans un environnement fortement contraint où la pose, l’éclairage, l’expression faciale et d’autres sources de variation sont sévèrement contrôlées. Cependant, ces approches sont souvent confinées aux domaines d’application limités parce que les environnements d’imagerie non-idéaux sont très fréquents dans les cas pratiques. Pour relever ces défis d’une manière adaptative, cette thèse porte sur le problème de reconnaissance faciale non contrôlée, dans lequel les images faciales présentent plus de variabilités sur les éclairages. Par ailleurs, une autre question essentielle vise à profiter des informations limitées de 3D pour collaborer avec les techniques basées sur 2D dans un système de reconnaissance faciale hétérogène. Pour traiter les diverses conditions d’éclairage, nous construisons explicitement un modèle de réflectance en caractérisant l’interaction entre la surface de la peau, les sources d’éclairage et le capteur de la caméra pour élaborer une explication de la couleur du visage. A partir de ce modèle basé sur la physique, une représentation robuste aux variations d’éclairage, à savoir Chromaticity Invariant Image (CII), est proposée pour la reconstruction des images faciales couleurs réalistes et sans ombre. De plus, ce processus de la suppression de l’ombre en niveaux de couleur peut être combiné avec les techniques existantes sur la normalisation d’éclairage en niveaux de gris pour améliorer davantage la performance de reconnaissance faciale. Les résultats expérimentaux sur les bases de données de test standard, CMU-PIE et FRGC Ver2.0, démontrent la capacité de généralisation et la robustesse de notre approche contre les variations d’éclairage. En outre, nous étudions l’usage efficace et créatif des données 3D pour la reconnaissance faciale hétérogène. Dans un tel scénario asymétrique, un enrôlement combiné est réalisé en 2D et 3D alors que les images de requête pour la reconnaissance sont toujours les images faciales en 2D. A cette fin, deux Réseaux de Neurones Convolutifs (Convolutional Neural Networks, CNN) sont construits. Le premier CNN est formé pour extraire les descripteurs discriminants d’images 2D/3D pour un appariement hétérogène. Le deuxième CNN combine une structure codeur-décodeur, à savoir U-Net, et Conditional Generative Adversarial Network (CGAN), pour reconstruire l’image faciale en profondeur à partir de son homologue dans l’espace 2D. Plus particulièrement, les images reconstruites en profondeur peuvent être également transmise au premier CNN pour la reconnaissance faciale en 3D, apportant un schéma de fusion qui est bénéfique pour la performance en reconnaissance. Notre approche a été évaluée sur la base de données 2D/3D de FRGC. Les expérimentations ont démontré que notre approche permet d’obtenir des résultats comparables à ceux de l’état de l’art et qu’une amélioration significative a pu être obtenue à l’aide du schéma de fusion
In recent years, biometrics have received substantial attention due to the evergrowing need for automatic individual authentication. Among various physiological biometric traits, face offers unmatched advantages over the others, such as fingerprints and iris, because it is natural, non-intrusive and easily understandable by humans. Nowadays conventional face recognition techniques have attained quasi-perfect performance in a highly constrained environment wherein poses, illuminations, expressions and other sources of variations are strictly controlled. However these approaches are always confined to restricted application fields because non-ideal imaging environments are frequently encountered in practical cases. To adaptively address these challenges, this dissertation focuses on this unconstrained face recognition problem, where face images exhibit more variability in illumination. Moreover, another major question is how to leverage limited 3D shape information to jointly work with 2D based techniques in a heterogeneous face recognition system. To deal with the problem of varying illuminations, we explicitly build the underlying reflectance model which characterizes interactions between skin surface, lighting source and camera sensor, and elaborate the formation of face color. With this physics-based image formation model involved, an illumination-robust representation, namely Chromaticity Invariant Image (CII), is proposed which can subsequently help reconstruct shadow-free and photo-realistic color face images. Due to the fact that this shadow removal process is achieved in color space, this approach could thus be combined with existing gray-scale level lighting normalization techniques to further improve face recognition performance. The experimental results on two benchmark databases, CMU-PIE and FRGC Ver2.0, demonstrate the generalization ability and robustness of our approach to lighting variations. We further explore the effective and creative use of 3D data in heterogeneous face recognition. In such a scenario, 3D face is merely available in the gallery set and not in the probe set, which one would encounter in real-world applications. Two Convolutional Neural Networks (CNN) are constructed for this purpose. The first CNN is trained to extract discriminative features of 2D/3D face images for direct heterogeneous comparison, while the second CNN combines an encoder-decoder structure, namely U-Net, and Conditional Generative Adversarial Network (CGAN) to reconstruct depth face image from its counterpart in 2D. Specifically, the recovered depth face images can be fed to the first CNN as well for 3D face recognition, leading to a fusion scheme which achieves gains in recognition performance. We have evaluated our approach extensively on the challenging FRGC 2D/3D benchmark database. The proposed method compares favorably to the state-of-the-art and show significant improvement with the fusion scheme

Orientador: Aparecido Nilceu Marana
Banca: Patricia Bellin Ribeiro
Banca: Roberta Spolon
Resumo: Na sociedade atual, a identificação precisa e rápida dos indivíduos é uma necessidade. Devido às crescentes preocupações sobre segurança, a Biometria tem sido proposta para este fim. A região ocular da face, que inclui o olho, as pálpebras, os cílios e as sobrancelhas, é uma das mais recentes modalidades biométricas sendo pesquisadas. Além da alta unicidade desta região da face, sua utilização representa um bom trade-off entre a utilização de toda a região da face e a utilização apenas da textura da íris dos olhos, pois possibilita uma gama maior de distâncias do indivíduo sendo identificado ao sensor. Este trabalho apresenta um novo método de autenticação de pessoas baseado em características oculares profundas, que são extraídas da região ocular da face usando uma CNN (Convolutional Neural Network). Em nosso método, em vez de usar diretamente os características profundas para a autenticação, usamos a diferença entre as características de referência e teste, gerando um vetor diferença. Então, nosso método adota uma estratégia de pares. Em seguida, um classificador SVM (Support Vector Machine) binário é treinado para determinar se um vetor diferença é genuíno ou impostor. O novo método proposto para autenticação de pessoas baseado em características oculares foi avaliado em diferentes bases de dados, contendo toda a face ou apenas a região ocular. Em nossos experimentos, a fusão de características oculares com características faciais obteve melhores resultados do que o uso...
Abstract: In modern society, accurate and quick identification of individuals is a necessity. Due to growing security concerns, Biometrics has been proposed for this purpose. The ocular region of the face, which includes the eye, eyelids, eyelashes and eyebrows, is one of the most recent biometric modalities being investigated. In addition to the high uniqueness of this region of the face, its use represents a good trade-off between the use of the entire face region and using only the texture of the iris of the eyes, since it allows a greater range of distances of the individual being identified to the sensor. This work presents a new method for identity authentication based on ocular deep features, which are extracted from the ocular region of the face by using a very deep CNN (Convolutional Neural Network). In our method, instead of using directly the deep features for the authentication, we use the difference between the probe and reference deep features, creating a difference vector. So, our method adopts a pairwise strategy. Then, a binary SVM (Support Vector Machine) classifier is trained to determine whether a given difference of deep features is genuine or impostor. The proposed new method for identity authentication based on ocular features was evaluated on different databases, containing the entire face or only the ocular region. In our experiments, the fusion of ocular features with facial features obtained better results than the use of features of the whole face when ...
Mestre

Dissertação (mestrado)—Universidade de Brasília, Instituto de Ciências Exatas, Departamento de Ciência da Computação, 2018.
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES).
Reconhecimento facial é uma área de grande interesse para governos e empresas nos dias de hoje, motivando diversas pesquisas na área. Atualmente, a maioria das pesquisas migrou para a utilização de aprendizado de máquina para apresentar soluções para o reconhecimento facial biométrico. As técnicas utilizando marcos faciais eram a principal abordagem para resolver reconhecimentos faciais biométricos e, mesmo com o avanço de novas técnicas, elas ainda são capazes de obter bons resultados em ambientes controlados. Porém, ainda existem problemas em aberto para serem solucionados, como o tratamento de gêmeos, variação de escala e o crescimento da face. Neste trabalho, inicialmente é feita uma pesquisa sobre os principais algoritmos de reconhecimento facial biométrico e, além disso, é proposto um novo método baseado em valores medidos (razões) de marcos faciais cefalométricos, que utilizam o tamanho da iris como um fator de normalização para solucionar a influência dos efeitos escala da face (crescimento facial) e melhorando os valores deEqual Error Rate (EER) para um sistema de reconhecimento facial em cenários específicos em 5%. Também foi feita uma análise para reduzir o número de razões necessárias, reduzindo das inicias 40 razões para 14, enquanto também aumentando o desempenho do método.
Nowadays, facial recognition arouses interest for governments and companies, motivating various reserches in the area. Presently, most researches have used techniques based on machine learning. Facial landmarks techniques were the first and main approach to solve biometric facial recognition and, even with the rise of newer techniques, they are still capable of achieving great results in controlled environments. However, there are still open problems to be solved, such as how to deal with twins, scale variation and the face growth. In this work, initially a research is made over the most important algorithms for biometric facial recognition and also, we propose a new method based on measured values (ratios) from facial cephalometric landmarks, which uses an iris size as a normalization factor to solve the influence of face scale (face growth) effect and improving Equal Error Rates (EER) scores for a facial recognition system in specifics scenarios under 5%. An analysis to reduce the number of ratios needed was also made, reducing from 40 to 14, while also increasing the performance of the method.

La reconnaissance automatique de visages est un domaine de recherche pour lequel un effort important a été consenti au cours des trois dernières décennies. Le present travail de thèse s'inscrit dans le cadre de l'un des thèmes d'actualité de ce domaine à savoir la reconnaissance faciale en 3D. Dans ce travail, nous nous sommes intéressés aux deux aspects complémentaires de ce sujet qui sont : la modélisation et la reconnaissance faciales tridimensionnelle. Alors que la modelisation a pour objectif l'acquisition de la forme 3D du visage, la reconnaissance vise l'identification d'un visage requête parmi des visages stockés dans une base de données ou bien la verification de son identité. Pour cela, deux approches été étudiées et mises en place : (i) une approche hybride d'acquisition faciale basée sur la stéréovision active et la modélisation géométrique, et (ii) une approche de recalage de surfaces faciales afin de mesurer les similarites entre les modèles 3D de visages. Une nouvelle base de donée incluant des acquisitions 3D, a été collectée dans le cadre du projet Technovision IV 2 afin d'effectuer des évaluations significatives sur les algorithmes developpés
Nowadays, face recognition represent one of the privileged fields of search due to the emergence of the security in many domains. This thesis lies within this scope, and more particularly, in the three-dimensional face recognition. In this work, we are interested to the complementary fields : 3D face modelling and recognition. Whereas modelling task aims at 3D face shape acquisition, recognition task aims at the identification of a probe face model among faces stored in a data base (gallery) or verify his identity. For that, two approaches are studied and implemented : (i) an hybrid approach for facial acquisition based on active vision and geometrical modelling, and (ii) an approach for aligning facial surfaces before computing similarities between 3D models. A new 3D face database is collected within the IV 2 French project in order to make signifiant experiments and evaluations of the developed algorithms

La littérature accorde un intérêt soutenu à l'étude des relations entre la morphologie basi-crânienne et les dysmorphies maxillo-faciales ou les malocclusions. Le but de ce travail est double : d'abord nous voulions analyser la forme basi-crânienne et ses rapports avec les schémas maxillo-faciaux ou les malocclusions. Ensuite nous voulions analyser la morphologie maxillo-faciale et ses relations avec les malocclusions. Nous avons utilisé les données scanner de 312 patients sélectionnés comme ayant des pathologies importantes " limite chirurgicale " et analysé ces donnée par deux méthodes : les procédés de la morphométrie géométrique et une analyse maxillo-faciale spécifique élaborée à Toulouse. Nous pouvons conclure à propos des deux principales questions : * La configuration basi-crânienne n'est pas significativement corrélée avec les différents types de malocclusion, elle reste très stable et elle ne joue aucun rôle étiologique dans le développement des malocclusions. * Au contraire, des schémas maxillo-faciaux spécifiques correspondant aux différentes malocclusions peuvent être décrits avec précision
There is an interest in the recent literature about the relationship between cranial base configuration and facial disharmonies or malocclusions, the conclusions of which are contradictory, due to small sample size and very poor methodology to appreciate cranial base shape. The aims of this work are double, the analyze of the cranial base configuration and its relationships with maxillo-facial schemes or malocclusion and the analyze of the maxillo-facial shape and its relationships with malocclusions. 312 patients selected with great pathologies "border line surgery" were used in this study and 3D method was applied, the morphometric geometry processes and a specific 3D maxillo-facial analysis elaborated in Toulouse. The results revealed two interesting finding, the basicranial configuration is not significantly correlated with types of malocclusion, it is remarkably stable, and it does not play any etiologic role in malocclusion appearance, and on the contrary, maxillofacial specific configurations, corresponding to different types of malocclusion, can be described precisely

In this doctoral dissertation, we principally explore the use of the temporal information available in video sequences for person and gender recognition; in particular, we focus on the analysis of head and facial motion, and their potential application as biometric identifiers. We also investigate how to exploit as much video information as possible for the automatic recognition; more precisely, we examine the possibility of integrating the head and mouth motion information with facial appearance into a multimodal biometric system, and we study the extraction of novel spatio-temporal facial features for recognition. We initially present a person recognition system that exploits the unconstrained head motion information, extracted by tracking a few facial landmarks in the image plane. In particular, we detail how each video sequence is firstly pre-processed by semi-automatically detecting the face, and then automatically tracking the facial landmarks over time using a template matching strategy. Then, we describe the geometrical normalisations of the extracted signals, the calculation of the feature vectors, and how these are successively used to estimate the client models through a Gaussian mixture model (GMM) approximation. In the end, we achieve person identification and verification by applying the probability theory and the Bayesian decision rule (also called Bayesian inference). Afterwards, we propose a multimodal extension of our person recognition system; more precisely, we successfully integrate the head motion information with mouth motion and facial appearance, by taking advantage of a unified probabilistic framework. In fact, we develop a new temporal subsystem that has an extended feature space enriched by some additional mouth parameters; at the same time, we introduce a complementary spatial subsystem based on a probabilistic extension of the original eigenface approach. In the end, we implement an integration step to combine the similarity scores of the two parallel subsystems, using a suitable opinion fusion (or score fusion) strategy. Finally, we investigate a practical method for extracting novel spatio-temporal facial features from video sequences, which are used to discriminate identity and gender. For this purpose we develop a recognition system called tomofaces, which applies the temporal X-ray transformation of a video sequence to summarise the facial motion and appearance information of a person into a single X-ray image. Then, we detail the linear projection from the X-ray image space to a low dimensional feature space, the estimation of the client models obtained by computing their cluster representatives, and the recognition of identity and gender through a nearest neighbour classifier using distances
Dans cette thèse, nous avons principalement exploré l'utilisation de l'information temporelle des séquences vidéo afin de l'appliquer à la reconnaissance de personne et de son genre; en particulier, nous nous concentrons sur l'analyse du mouvement de la tête et du visage ainsi que sur leurs applications potentielles comme éléments d'identification biométriques. De plus, nous cherchons à exploiter la majorité de l'information contenue dans la vidéo pour la reconnaissance automatique; plus précisément, nous regardons la possibilité d'intégrer dans un système biométrique multimodal l'information liée au mouvement de la tête et de la bouche avec celle de l'aspect du visage, et nous étudions l'extraction des nouveaux paramètres spatio-temporels pour la reconnaissance faciale. Nous présentons d'abord un système de reconnaissance de la personne qui exploite l'information relative au mouvement spontané de la tête. Cette information est extraite par le suivi dans le plan image de certains éléments caractéristiques du visage. En particulier, nous détaillons la façon dont dans chaque séquence vidéo le visage est tout d'abord détecté semi-automatiquement, puis le suivi automatique dans le temps de certains éléments caractéristiques en utilisant une approche basée sur l'appariement de bloques (template matching). Ensuite, nous exposons les normalisations géométriques des signaux que nous avons obtenus, le calcul des vecteurs caractéristiques, et la façon dont ils sont utilisés pour estimer les modèles des clients, approximés avec des modèles de mélange de gaussiennes. En fin de compte, nous parvenons à identifier et vérifier l'identité de la personne en appliquant la théorie des probabilités et la règle de décision bayésienne (aussi appelée inférence bayésienne). Nous proposons ensuite une extension multimodale de notre système de reconnaissance de la personne; plus précisément, nous intégrons à travers un cadre probabiliste unifié l'information sur le mouvement de la tête avec celles liées au mouvement de la bouche et à l'aspect du visage. En fait nous développons un nouveau sous-système temporel qui a un espace caractéristique étendu, lequel est enrichi par certains paramètres supplémentaires relatif au mouvement de la bouche; dans le même temps nous introduisons un sous-système spatial complémentaire au précédent, basé sur une extension probabiliste de l'approche Eigenfaces d'origine. Ensuite, une étape d'intégration combine les scores de similarité des deux sous-systèmes parallèles, grâce à une stratégie appropriée de fusion d'opinions. Enfin nous étudions une méthode pratique pour extraire de nouveaux paramètres spatio-temporels liés au visage à partir des séquences vidéo; le but est de distinguer l'identité et le genre de la personne. À cette fin nous développons un système de reconnaissance appelé tomovisages (tomofaces), qui applique la technique de la tomographie vidéo pour résumer en une seule image l'information relative au mouvement et à l'aspect du visage d'une personne. Puis, nous détaillons la projection linéaire à partir de l'espace de l'image en rayons X à un espace caractéristique de dimension réduite, l'estimation des modèles des utilisateurs en calculant les représentants des clusters correspondants, et la reconnaissance de l'identité et du genre par le biais d'un classificateur de plus proche voisin, qui adopte des distances dans le sous-espace
In questa tesi di dottorato esploriamo la possibilità di riconoscere l'identità e il sesso di una persona attraverso l'utilizzo dell'informazione temporale disponibile in alcune sequenze video, in particolare ci concentriamo sull'analisi del movimento della testa e del viso, nonché del loro potenziale utilizzo come identificatiori biometrici. Esaminiamo inoltre la problematica relativa al fatto di sfruttare la maggior parte dell'informazione presente nei video per effettuare il riconoscimento automatico della persona; più precisamente, analizziamo la possibilità di integrare in un sistema biometrico multimodale l'informazione relativa al movimento della testa e della bocca con quella dell'aspetto del viso, e studiamo il calcolo di nuovi parametri spazio-temporali che siano utilizzabili per il riconoscimento stesso. In primo luogo presentiamo un sistema di riconoscimento biometrico della persona che sfrutti l'informazione legata al movimento naturale della testa, il quale è estratto seguendo la posizione nel piano immagine di alcuni elementi caratteristici del viso. In particolare descriviamo come in una sequenza video il volto venga dapprima individuato semiautomaticamente, e come poi alcuni suoi elementi caratteristici siano localizzati nel tempo tramite un algoritmo automatico di messa in corrispondenza di modelli (template matching) permettendo di seguirne la posizione. Spieghiamo quindi le normalizzazioni geometriche dei segnali che abbiamo ricavato, il calcolo dei vettori caratteristici, ed il modo in cui questi sono utilizzati per stimare i modelli degli utilizzatori, approssimandoli tramite delle misture di distribuzioni gaussiane (Gaussian mixture models). Alla fine otteniamo l'identificazione e la verifica dell'identità della persona applicando la teoria delle probabilità e la regola di decisione o inferenza bayesiana. In seguito proponiamo un'estensione multimodale del nostro sistema di riconoscimento della persona; più precisamente, tramite un approccio probabilistico unificato, integriamo l'informazione sul movimento della testa con quelle relative al movimento della bocca e all'aspetto del viso. Infatti sviluppiamo un nuovo sottosistema temporale che possiede uno spazio caratteristico esteso, arricchito di alcuni parametri aggiuntivi legati al movimento della bocca; contemporaneamente, introduciamo un sottosistema spaziale complementare al precedente, basato su un'estensione probabilistica dell'approccio Eigenfaces originale. Alla fine implementiamo uno stadio di fusione, che metta insieme i valori di somiglianza dei due sottosistemi paralleli, attraverso un'appropriata strategia di fusione delle opinioni. Infine investighiamo un metodo pratico per estrarre nuovi parametri spazio-temporali relativi al volto a partire da sequenze video, i quali sono utilizzati per distinguere l'identità ed il sesso della persona. A questo riguardo sviluppiamo un sistema di riconoscimento chiamato tomovolti (tomofaces), il quale utilizza la tecnica della tomografia video per riassumere in una sola immagine l'informazione relativa all'aspetto ed al movimento del volto di una persona. Poi descriviamo la proiezione lineare dallo spazio dell'immagine ai raggi X ad un spazio caratteristico di dimensione ridotta, la stima dei modelli degli utilizzatori attraverso il calcolo dei rappresentanti corrispondenti ad ogni cluster, ed il riconoscimento dell'identità e del genere attraverso un classificatore al vicino più prossimo (nearest neighbour classifier), che adopera le distanze nel sottospazio

Made available in DSpace on 2014-06-11T19:29:40Z (GMT). No. of bitstreams: 0 Previous issue date: 2011-08-05Bitstream added on 2014-06-13T19:38:57Z : No. of bitstreams: 1 kami_gjc_me_sjrp.pdf: 1342570 bytes, checksum: 240c6d6b92fda1861dfbed94c9213a10 (MD5)
As técnicas para identificação biométrica têm evoluído cada vez mais devido à necessidade que os seres humanos têm de identificar as pessoas em tempo real e de forma precisa para permitir o acesso a determinados recursos, como por exemplo, as aplicações e serviços WEB. O reconhecimento facial é uma técnica biométrica que apresenta várias vantagens em relação às demais, tais como: uso de equipamentos simples e baratos para a obtenção das amostras e a possibilidade de se realizar o reconhecimento em sigilo e à distância. O reconhecimento de faces a partir de vídeo é uma tendência recente na área de Biometria. Esta dissertação tem por objetivo principal comparar diferentes técnicas de reconhecimento facial a partir de vídeo para determinar as que apresentam um melhor compromisso entre tempo de processamento e precisão. Outro objetivo é a incorporação dessas melhores técnicas no sistema de autenticação biométrica em ambientes de E-Learning, proposto em um trabalho anterior. Foi comparado o classificador vizinho mais próximo usando as medidas de distância Euclidiana e Mahalanobis com os seguintes classificadores: Redes Neurais MLP e SOM, K Vizinhos mais Próximos, Classificador Bayesiano, Máquinas de Vetores de Suporte (SVM) e Floresta de Caminhos Ótimos (OPF). Também foi avaliada a técnica de Modelos Ocultos de Markov (HMM). Nos experimentos realizados com a base Recogna Video Database, criada especialmente para uso neste trabalho, e Honda/UCSD Video Database, os classificadores apresentaram os melhores resultados em termos de precisão, com destaque para o classificador SVM da biblioteca SVM Torch. A técnica HMM, que incorpora informações temporais, apresentou resultados melhores do que as funções de distância, em termos de precisão, mas inferiores aos classificadores
The biometric identification techniques have evolved increasingly due to the need that humans have to identify people in real time to allow access to certain resources, such as applications and Web services. Facial recognition is a biometric technique that has several advantages over others. Some of these advantages are the use of simple and cheap equipment to obtain the samples and the ability to perform the recognition in covert mode. The face recognition from video is a recent approach in the area of Biometrics. The work in this dissertation aims at comparing different techniques for face recognition from video in order to find the best rates on processing time and accuracy. Another goal is the incorporation of these techniques in the biometric authentication system for E-Learning environments, proposed in an earlier work. We have compared the nearest neighbor classifier using the Euclidean and Mahalanobis distance measures with some other classifiers, such as neural networks (MLP and SOM), k-nearest neighbor, Bayesian classifier, Support Vector Machines (SVM), and Optimum Path Forest (OPF). We have also evaluated the Hidden Markov Model (HMM) approach, as a way of using the temporal information. In the experiments with Recogna Video Database, created especially for this study, and Honda/UCSD Video Database, the classifiers obtained the best accuracy, especially the SVM classifier from the SVM Torch library. HMM, which takes into account temporal information, presented better performance than the distance metrics, but worse than the classifiers

The face is one of the best biometrics for person identification and verification related applications, because it is natural, non-intrusive, and socially weIl accepted. Unfortunately, an human faces are similar to each other and hence offer low distinctiveness as compared with other biometrics, e.g., fingerprints and irises. Furthermore, when employing facial texture images, intra-class variations due to factors as diverse as illumination and pose changes are usually greater than inter-class ones, making 2D face recognition far from reliable in the real condition. Recently, 3D face data have been extensively investigated by the research community to deal with the unsolved issues in 2D face recognition, Le., illumination and pose changes. This Ph.D thesis is dedicated to robust face recognition based on three dimensional data, including only 3D shape based face recognition, textured 3D face recognition as well as asymmetric 3D-2D face recognition. In only 3D shape-based face recognition, since 3D face data, such as facial pointclouds and facial scans, are theoretically insensitive to lighting variations and generally allow easy pose correction using an ICP-based registration step, the key problem mainly lies in how to represent 3D facial surfaces accurately and achieve matching that is robust to facial expression changes. In this thesis, we design an effective and efficient approach in only 3D shape based face recognition. For facial description, we propose a novel geometric representation based on extended Local Binary Pattern (eLBP) depth maps, and it can comprehensively describe local geometry changes of 3D facial surfaces; while a 81FT -based local matching process further improved by facial component and configuration constraints is proposed to associate keypoints between corresponding facial representations of different facial scans belonging to the same subject. Evaluated on the FRGC v2.0 and Gavab databases, the proposed approach proves its effectiveness. Furthermore, due tq the use of local matching, it does not require registration for nearly frontal facial scans and only needs a coarse alignment for the ones with severe pose variations, in contrast to most of the related tasks that are based on a time-consuming fine registration step. Considering that most of the current 3D imaging systems deliver 3D face models along with their aligned texture counterpart, a major trend in the literature is to adopt both the 3D shape and 2D texture based modalities, arguing that the joint use of both clues can generally provides more accurate and robust performance than utilizing only either of the single modality. Two important factors in this issue are facial representation on both types of data as well as result fusion. In this thesis, we propose a biological vision-based facial representation, named Oriented Gradient Maps (OGMs), which can be applied to both facial range and texture images. The OGMs simulate the response of complex neurons to gradient information within a given neighborhood and have properties of being highly distinctive and robust to affine illumination and geometric transformations. The previously proposed matching process is then adopted to calculate similarity measurements between probe and gallery faces. Because the biological vision-based facial representation produces an OGM for each quantized orientation of facial range and texture images, we finally use a score level fusion strategy that optimizes weights by a genetic algorithm in a learning pro cess. The experimental results achieved on the FRGC v2.0 and 3DTEC datasets display the effectiveness of the proposed biological vision-based facial description and the optimized weighted sum fusion. [...]

Orientador: Aparecido Nilceu Marana
Banca: Hélio Pedrini
Banca: Aledir Silveira Pereira
Resumo: As técnicas para identificação biométrica têm evoluído cada vez mais devido à necessidade que os seres humanos têm de identificar as pessoas em tempo real e de forma precisa para permitir o acesso a determinados recursos, como por exemplo, as aplicações e serviços WEB. O reconhecimento facial é uma técnica biométrica que apresenta várias vantagens em relação às demais, tais como: uso de equipamentos simples e baratos para a obtenção das amostras e a possibilidade de se realizar o reconhecimento em sigilo e à distância. O reconhecimento de faces a partir de vídeo é uma tendência recente na área de Biometria. Esta dissertação tem por objetivo principal comparar diferentes técnicas de reconhecimento facial a partir de vídeo para determinar as que apresentam um melhor compromisso entre tempo de processamento e precisão. Outro objetivo é a incorporação dessas melhores técnicas no sistema de autenticação biométrica em ambientes de E-Learning, proposto em um trabalho anterior. Foi comparado o classificador vizinho mais próximo usando as medidas de distância Euclidiana e Mahalanobis com os seguintes classificadores: Redes Neurais MLP e SOM, K Vizinhos mais Próximos, Classificador Bayesiano, Máquinas de Vetores de Suporte (SVM) e Floresta de Caminhos Ótimos (OPF). Também foi avaliada a técnica de Modelos Ocultos de Markov (HMM). Nos experimentos realizados com a base Recogna Video Database, criada especialmente para uso neste trabalho, e Honda/UCSD Video Database, os classificadores apresentaram os melhores resultados em termos de precisão, com destaque para o classificador SVM da biblioteca SVM Torch. A técnica HMM, que incorpora informações temporais, apresentou resultados melhores do que as funções de distância, em termos de precisão, mas inferiores aos classificadores
Abstract: The biometric identification techniques have evolved increasingly due to the need that humans have to identify people in real time to allow access to certain resources, such as applications and Web services. Facial recognition is a biometric technique that has several advantages over others. Some of these advantages are the use of simple and cheap equipment to obtain the samples and the ability to perform the recognition in covert mode. The face recognition from video is a recent approach in the area of Biometrics. The work in this dissertation aims at comparing different techniques for face recognition from video in order to find the best rates on processing time and accuracy. Another goal is the incorporation of these techniques in the biometric authentication system for E-Learning environments, proposed in an earlier work. We have compared the nearest neighbor classifier using the Euclidean and Mahalanobis distance measures with some other classifiers, such as neural networks (MLP and SOM), k-nearest neighbor, Bayesian classifier, Support Vector Machines (SVM), and Optimum Path Forest (OPF). We have also evaluated the Hidden Markov Model (HMM) approach, as a way of using the temporal information. In the experiments with Recogna Video Database, created especially for this study, and Honda/UCSD Video Database, the classifiers obtained the best accuracy, especially the SVM classifier from the SVM Torch library. HMM, which takes into account temporal information, presented better performance than the distance metrics, but worse than the classifiers
Mestre

L’étude menée dans le cadre de cette thèse vise l’étude du rôle de la dynamique de formes faciales 3D à révéler l’identité des personnes et leurs états émotionnels. Pour se faire, nous avons proposé un cadre géométrique pour l’étude des formes faciales 3D et leurs dynamiques dans le temps. Une séquence 3D est d’abord divisée en courtes sous-séquences, puis chacune des sous-séquences obtenues est représentée dans une variété de Grassmann (ensemble des sous-espaces linéaires de dimension fixe). Nous avons exploité la géométrie de ces variétés pour comparer des sous-séquences 3D, calculer des statistiques (telles que des moyennes) et quantifier la divergence entre des éléments d’une même variété Grassmannienne. Nous avons aussi proposé deux représentations possibles pour les deux applications cibles – (1) la première est basée sur les dictionnaires (de sous-espaces) associée à des techniques de Dictionary Learning Sparse Coding pour la reconnaissance d’identité et (2) le représentation par des trajectoires paramétrées par le temps sur les Grassmanniennes couplée avec une variante de l’algorithme de classification SVM, permettant un apprentissage avec des données partielles, pour la détection précoce des émotions spontanée. Les expérimentations réalisées sur les bases publiques BU-4DFE, Cam3D et BP4D-Spontaneous montrent à la fois l’intérêt du cadre géométrique proposé (en terme de temps de calcul et de robustesse au bruit et aux données manquantes) et les représentations adoptées (dictionnaires pour la reconnaissance d’identité et trajectoires pour la détection précoce des émotions spontanées)
In this thesis, we have investigated the problems of identity recognition and emotion detection from facial 3D shapes animations (called 4D faces). In particular, we have studied the role of facial (shapes) dynamics in revealing the human identity and their exhibited spontaneous emotion. To this end, we have adopted a comprehensive geometric framework for the purpose of analyzing 3D faces and their dynamics across time. That is, a sequence of 3D faces is first split to an indexed collection of short-term sub-sequences that are represented as matrix (subspace) which define a special matrix manifold called, Grassmann manifold (set of k-dimensional linear subspaces). The geometry of the underlying space is used to effectively compare the 3D sub-sequences, compute statistical summaries (e.g. sample mean, etc.) and quantify densely the divergence between subspaces. Two different representations have been proposed to address the problems of face recognition and emotion detection. They are respectively (1) a dictionary (of subspaces) representation associated to Dictionary Learning and Sparse Coding techniques and (2) a time-parameterized curve (trajectory) representation on the underlying space associated with the Structured-Output SVM classifier for early emotion detection. Experimental evaluations conducted on publicly available BU-4DFE, BU4D-Spontaneous and Cam3D Kinect datasets illustrate the effectiveness of these representations and the algorithmic solutions for identity recognition and emotion detection proposed in this thesis

Made available in DSpace on 2015-05-14T12:36:38Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 2501696 bytes, checksum: a1f65e686060aa68aa5e43dee1e6c0b5 (MD5) Previous issue date: 2012-07-30
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
Reliable identification systems have become key components in many applications that provide services to authenticated users. Since traditional authentication methods (such as using passwords or smartcards) can be manipulated in order to bypass the systems, biometric authentication methods have been receiving more attention in recent years. One of the biometric traits is the face. The problem of recognizing faces in video and photo still is an object of research, since there are many factors that influence the detection and recognition, such as lighting, position of the face, the background image, different facial expressions, etc. One can perform face recognition using Discrete Cosine Transform (DCT). In order to adjust a face recognition system to uncontrolled environments, two improvements for it were developed in this work: a image normalization module with respect to rotation and scale, and a change in the feature extraction module through the insertion of a non-ideal low-pass filter. The system and its modifications were tested on the following face databases: UFPB, ORL, Yale, and VSoft GTAV, developed specially for the job. Tests showed the efficiency of the image normalization module, but the system still is not adequate for every environment.
Sistemas de identificação confiáveis tornaram-se componentes chaves de várias aplicações que disponibilizam serviços para usuários autenticados. Uma vez que métodos de autenticação tradicionais (como os que utilizam senhas ou smartcards) podem ser manipulados com o objetivo de burlar os sistemas, métodos de autenticação biométrica vêm recebendo mais atenção nos últimos anos. Um dos traços biométricos é a face. O problema do reconhecimento de faces em vídeo e foto é objeto de pesquisa, uma vez que existem muitos fatores que influenciam na detecção e no reconhecimento, tais como: iluminação, posição da face, imagem ao fundo, diferentes expressões faciais, etc. É possível realizar reconhecimento facial utilizando a Transformada Cosseno Discreta (DCT). Com o intuito de adequar um Sistema de Detecção e Reconhecimento de Faces a ambientes não controlados, neste trabalho foram desenvolvidas duas melhorias para ele: um módulo normalizador de imagens em relação à rotação e à escala e uma modificação na etapa de seleção de atributos, por meio da inserção de um filtro passa-baixas não ideal. O sistema e suas modificações foram testados nos bancos de faces UFPB, ORL, Yale, GTAV e Vsoft, desenvolvido especialmente para o trabalho. Os testes mostraram a eficácia do módulo de normalização da imagem, mas ainda assim o sistema não é adequado para qualquer ambiente.

Cette thèse a pour objectif de modéliser, par approches biométriques, l’évolution dans le temps du visage humain, en partant de l’âge enfant, jusqu’à un âge adulte. Ces travaux sur le vieillissement rentrent dans le cadre des activités de recherche du groupe biométrie du laboratoire LiSSi (UPEC).Comme il est connu, l’évolution des traits dues au vieillissement dépend deplusieurs facteurs intrinsèques ou extrinsèques, dont : la génétique, l’origine ethnique, le mode de vie, etc. En considérons les modèles paramétriques proposés dans cette thèse, nous exploitons entre autres, les similitudes des caractéristiques extraites chez des individus d’une même catégorie d’âge. Ces similitudes sont intégrées dans nos modèles afin de pouvoir estimer l’apparence faciale à un âge spécifique. Contrairement aux nombreuses études traitant les modèles prédictifs de vieillissement facial, cette thèse propose pour la première fois un modèle réversible permettant également le rajeunissement numérique de l’apparence du visage que nous appellerons, modèle de prédiction arrière d’apparence. Quant à la prédiction avant, notre contribution s’est orientée vers la proposition d’un modèle non-linaire paramétrique de vieillissement permettant de prendre en considération les facteurs accélérateurs de vieillissements liés au mode de vie des individus. De manière générale, nous nous sommes intéressés aux conséquences de certaines addictions de type (drogues, alcool,exposition au soleil, etc.), sur le vieillissement prématuré du visage. Par conséquent,nous avons proposé des modèles sensibles à certains de ces facteurs en se basant sur des analyses statistiques. Comme retombés socio-économiques, cette étude a pour objectif de sensibiliser les jeunes personnes par rapport aux dangers liés à la consommation excessives de certaines substances, voire à l’addiction à certaines pratiques.Les études que nous avons menées durant cette thèse, ont nécessité la constitution d’une base de données contenant plus de 1600 images faciales. Cette base de données a permis le développement 30 modèles de visages «Face Templates». Suite à cela, nous avons créé une base de données d’évaluation, appelée «Face Time-Machine (FaceTiM)». Constituée à partir de 120 sujets, cette base de données est mise à disposition des chercheurs afin qu’ils puissent reproduire les résultats que nous avons obtenus, évaluer les performances, et enfin contribuer à l’amélioration des modèles proposés
The main focus of this thesis is to model the evolution trajectory of human face from infancy to senility using the biometrics facial features.The manifestation of facial changes caused by ageing depends on different factors such as genetic, ethnicity and lifestyle. Nevertheless, individuals in the same age group share some facial similarities. These resemblances can be employed to approximate the facial appearance of an individual in the bygone or the forthcoming years.Unlike numerous studies dealing with predictive face ageing models, for the first time, this thesis proposes the first Backward Facial Ageing Model aiming at digitally rejuvenate an adult face appearance down to its early childhood. We also present the Forward Facial Ageing Model to predict the adult face appearance in its future by taking into account the naturalageing trajectory. The main purpose of Forward Facial Ageing Model is to have a base model for the supplementary ageing models such as behavioural models.In this thesis for the first time in face ageing studies, the effects of different lifestyle behaviours are integrated into the facial ageing models. The Behavioural Facial Ageing Models predict the feature of a young face in case of having the high-risk lifestyle habits. The main attempt of these models is to illustrate the adverse effects of unsafe lifestyle behaviourson the senility of the face, aiming to prevent the youth from becoming involved in these habits. The Facial Ageing Modeling Database, contains over 1600 facial images, is collected to construct the models and 30 Face Templates for the purpose of the face ageing studies.Besides, the Face Time-Machine Database from 120 subjects is created and published to testand evaluate the results. For the proposed approach face contour and different components are modified non-linearly based on an estimated geometrical model related to the trajectory of growth or ageing. Moreover, the face texture is adapted by mapping a Face Template to the estimated geometrical model. Then, the effects of each lifestyle habit are set up to the primal predictive model.The evaluations of the results indicate that the proposed models are remarkably accurate to estimate the correct face appearance of an individual in the target age. While the simulated facial images are realistic and have the appearance, geometrical and textural characteristics of the target age, the personal identity and details of the input face images are preserved

Orientadores: Alexandre Xavier Falcão, Giovani Chiachia
Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Computação
Made available in DSpace on 2018-08-27T14:41:07Z (GMT). No. of bitstreams: 1 Martins_SamuelBotter_M.pdf: 4782261 bytes, checksum: 63cd58756e3fe70ffe625d42974b1a78 (MD5) Previous issue date: 2015
Resumo: Sistemas automáticos de reconhecimento de faces tem atraído a atenção da indústria e da academia, devido à gama de possíveis aplicações, tais como vigilância, controle de acesso, etc. O recente progresso em tais sistemas motiva o uso de técnicas de aprendizado em profundidade e classificadores específicos para cada usuário em cenários de operação não-controlado, que apresentam variações consideráveis em pose, iluminação, etc. Sistemas automáticos de reconhecimento de faces possibilitam construir bases de imagens anotadas por meio do processo de cadastramento de novos usuários. Porém, à medida que as bases de dados crescem, torna-se crucial reduzir o número de amostras negativas usadas para treinar classificadores específicos para cada usuário, devido às limitações de processamento e tempo de resposta. Tal processo de aprendizado discriminativo durante o cadastramento de novos indivíduos tem implicações no projeto de sistemas de reconhecimento de faces. Apesar deste processo poder aumentar o desempenho do reconhecimento, ele também pode afetar a velocidade do cadastramento, prejudicando, assim, a experiência do usuário. Neste cenário, é importante selecionar as amostras mais informativas buscando maximizar o desempenho do classificador. Este trabalho resolve tal problema propondo um método de aprendizado discriminativo durante o cadastramento de usuários com o objetivo de não afetar a velocidade e a confiabilidade do processo. Nossa solução combina representações de alta dimensão com um algoritmo que rapidamente minera imagens faciais negativas de um conjunto de minerção grande para assim construir um classificador específico para cada usuário, baseado em máquinas de vetores de suporte. O algoritmo mostrou ser robusto em construir pequenos e eficazes conjuntos de treinamento com as amostras negativas mais informativas para cada indivíduo. Avaliamos nosso método em duas bases contendo imagens de faces obtidas no cenário de operação não-controlado, chamadas PubFig83 e Mobio, e mostramos que nossa abordagem é capaz de alcançar um desempenho superior em tempos interativos, quando comparada com outras cinco abordagens consideradas. Os resultados indicam que o nosso método tem potencial para ser explorado pela indústria com mínimo impacto na experiência do usuário. Além disso, o algoritmo é independente de aplicação, podendo ser uma contribuição relevante para sistemas biométricos que visam manter a robustez à medida que o número de usuários aumenta
Abstract: Automatic face recognition has attracted considerable attention from the industry and academy due to its wide range of applications, such as video surveillance, access control, online transactions, suspect identification, etc. The recent progress in face recognition systems motivates the use of deep learning techniques and user-specific face representation and classification models for unconstrained scenarios, which present considerable variations in pose, face appearance, illumination, etc. Automatic face recognition systems make possible to build annotated face datasets through user enrollment. However, as the face datasets grow, it becomes crucial to reduce the number of negative samples used to train user-specific classifiers, due to processing constraints and responsiveness. Such a discriminative learning process during the enrollment of new individuals has implications in the design of face recognition systems. Even though it might increase recognition performance, it may affect the speed of the enrollment, which in turn may affect the user experience. In this scenario, it is important to select the most informative samples in order to maximize the performance of the classifier. This work addresses this problem by proposing a discriminative learning method during user enrollment with the challenges of not negatively affecting the speed and reliability of the process, and so the user experience. Our solution combines high-dimensional representations from deep learning with an algorithm for rapidly mining negative face images from a large mining set to build an effective classification model based on linear support vector machines for each specific user. The negative mining algorithm has shown to be robust in building small and effective training sets with the most informative negative samples for each given individual. We evaluate our approach on two unconstrained datasets, namely PubFig83 and Mobio, and show that it is able to attain superior performance, within interactive response times, as compared to five other baseline approaches that use the same classification scheme. The results indicate that our approach has potential to be exploited by the industry with minimum impact to the user experience. Moreover, the algorithm is application-independent. Hence, it may be a relevant contribution for biometric systems that aim to maintain robustness as the number of users increases
Mestrado
Ciência da Computação
Mestre em Ciência da Computação

La identificación de personas mediante la tecnología biométrica ha revolucionado la estructura y flujo de trabajo en lo público y privado, entre ellos los servicios de identificación policial, gabinetes de policía científica o laboratorios de criminalística, que han trascurrido de los métodos físicos y mecánicos a los sistemas automatizados y digitalizados, haciendo dinámicos los servicios de análisis de evidencias de huellas dactilares e imágenes faciales que antes demoraban mucho tiempo por la búsqueda comparativa y manual en un gran archivo. Hoy, la tecnología biométrica lo realiza en segundos, buscando en una base de datos que puede contener miles o millones de registros. Sin embargo, la optimización de la función de identificación criminalística trae consigo el incremento de requerimientos para análisis de evidencias por parte de los operadores de justicia. Esto da como consecuencia de la elevada incidencia de hechos delictivos de distinta tipología, que requiere la identificación del presunto autor o víctima para la continuación del proceso de investigación criminal, la administración de justicia y satisfacción de la demanda de los ciudadanos que acuden a los organismos jurisdiccionales en busca de un resarcimiento o reparación de su daño. Este es el caso de la División de Identificación de la Dirección de Criminalística de la Policía Nacional del Perú (PNP), que inició su labor especializada con apoyo de la tecnología biométrica de identificación de huellas dactilares denominada AFIS (Automated Fingerprint Identification System) desde el año 2012 y la incorporación del módulo de reconocimiento facial MFI (Morpho Face Identification) desde el año 2015. Esta ha visto rebasada su capacidad resolutiva, acumulando casos pendientes sin resolver por identificación de huellas dactilares e imágenes faciales. Mediante la presente investigación se ha logrado identificar diferentes variables (causas) que determinan los efectos que traen como consecuencia, la ineficacia de capacidad resolutiva de la Dirección de Identificación Criminalística de la PNP. Para dar respuesta a esta problemática, se ha formulado una propuesta de solución para atender la más relevante y viable, identificada como la escasez del personal especializado, con lo cual se pretende contribuir a la solución del problema o parte del mismo. Es necesario señalar que existen otras variables más complejas de atender, que también influyen en el problema y son al igual motivo de investigación para su atención y propuesta de solución.
The identification of people using biometric technology has revolutionized the structure and workflow of public and private entities, such as the case of forensic police identification services, scientific police offices or crime laboratories, which have evolved from physical methods and mechanical to the automated and digitized systems, making dynamic the services of analysis of evidence of fingerprints and facial images that used to take a long time for comparative and manual searching in a large archive; Today, biometric technology does it in seconds, searching a database that can contain thousands or millions of records. However, the optimization of the criminalistic identification function also brings with it the increase in requirements for the analysis of evidence by justice operators, even more so if there is a high incidence of criminal acts of different types, which requires identification. Of the alleged perpetrator or victim for the continuation of the criminal investigation process, the administration of justice and satisfaction of the demand of citizens who go to the jurisdictional bodies in search of compensation or reparation for their damage. This is the case of the Identification Division of the Criminalistics Directorate of the National Police of Peru (PNP), which began its specialized work with the support of the biometric fingerprint identification technology called AFIS (Automated Fingerprint Identification System) from the year 2012 and the incorporation of the MFI (Morpho Face Identification) facial recognition module since 2015; which has been exceeded its ability to resolve, accumulating pending cases unsolved by identification of fingerprints and facial images. Through the present investigation, it has been possible to identify different variables (causes) that determine the effects that, as a consequence, the ineffectiveness of the resolving capacity of the PNP Criminalistics Identification Directorate, formulating a solution proposal to attend to the most relevant and viable one, identified as the shortage of specialized personnel, with which it is intended to contribute to the solution of the problem or part of it, considering that there are other more complex variables to address, which also influence the problem and are the same reason for research for their care and proposed solution.

Orientadores: Alexandre Xavier Falcão, Anderson de Rezende Rocha
Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Computação
Made available in DSpace on 2018-08-23T15:41:33Z (GMT). No. of bitstreams: 1 Chiachia_Giovani_D.pdf: 4376963 bytes, checksum: 8f7d18d591f2a5d943313d89416f96d4 (MD5) Previous issue date: 2013
Resumo: Os seres humanos são especialistas natos em reconhecimento de faces, com habilidades que excedem em muito as dos métodos automatizados vigentes, especialmente em cenários não controlados, onde não há a necessidade de colaboração por parte do indivíduo sendo reconhecido. No entanto, uma característica marcante do reconhecimento de face humano é que nós somos substancialmente melhores no reconhecimento de faces familiares, provavelmente porque somos capazes de consolidar uma grande quantidade de experiência prévia com a aparência de certo indivíduo e de fazer uso efetivo dessa experiência para nos ajudar no reconhecimento futuro. De fato, pesquisadores em psicologia têm até mesmo sugeridos que a representação interna que fazemos das faces pode ser parcialmente adaptada ou otimizada para rostos familiares. Enquanto isso, a situação análoga no reconhecimento facial automatizado | onde um grande número de exemplos de treinamento de um indivíduo está disponível | tem sido muito pouco explorada, apesar da crescente relevância dessa abordagem na era das mídias sociais. Inspirados nessas observações, nesta tese propomos uma abordagem em que a representação da face de cada pessoa é explicitamente adaptada e realçada com o intuito de reconhecê-la melhor. Apresentamos uma coleção de métodos de aprendizado que endereça e progressivamente justifica tal abordagem. Ao aprender e operar com representações específicas para face de cada pessoa, nós somos capazes de consistentemente melhorar o poder de reconhecimento dos nossos algoritmos. Em particular, nós obtemos resultados no estado da arte na base de dados PubFig83, uma desafiadora coleção de imagens instituída e tornada pública com o objetivo de promover o estudo do reconhecimento de faces familiares. Nós sugerimos que o aprendizado de representações específicas para face de cada pessoa introduz uma forma intermediária de regularização ao problema de aprendizado, permitindo que os classificadores generalizem melhor através do uso de menos |, porém mais relevantes | características faciais
Abstract: Humans are natural face recognition experts, far outperforming current automated face recognition algorithms, especially in naturalistic, \in-the-wild" settings. However, a striking feature of human face recognition is that we are dramatically better at recognizing highly familiar faces, presumably because we can leverage large amounts of past experience with the appearance of an individual to aid future recognition. Researchers in psychology have even suggested that face representations might be partially tailored or optimized for familiar faces. Meanwhile, the analogous situation in automated face recognition, where a large number of training examples of an individual are available, has been largely underexplored, in spite of the increasing relevance of this setting in the age of social media. Inspired by these observations, we propose to explicitly learn enhanced face representations on a per-individual basis, and we present a collection of methods enabling this approach and progressively justifying our claim. By learning and operating within person-specific representations of faces, we are able to consistently improve performance on both the constrained and the unconstrained face recognition scenarios. In particular, we achieve state-of-the-art performance on the challenging PubFig83 familiar face recognition benchmark. We suggest that such person-specific representations introduce an intermediate form of regularization to the problem, allowing the classifiers to generalize better through the use of fewer | but more relevant | face features
Doutorado
Ciência da Computação
Doutor em Ciência da Computação

Made available in DSpace on 2015-05-14T12:36:31Z (GMT). No. of bitstreams: 1 parte1.pdf: 2637857 bytes, checksum: 9c7d375174ead2b03a7b858b1d087f4d (MD5) Previous issue date: 2009-08-13
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES
The increased demand for Distance Education and Virtual Learning Environment (VLE) appearance and development are in a scenario where minimizing distances and boosting tracking of students are aspired. However, current VLE don t have the necessary resources to identify the users in the development of their activities. This dissertation presents an online face identification system for use in VLE that has client-server architecture and combines stages of face recognition processing to modules of a mechanism monitor remote users via a webcam. To accomplish the face recognition was used a approach to select coefficients of Discrete Cosine Transform (DCT), which it obtained 97.12% rate of correct answers in a database of images collected without normalization of faces. This work describes, in addition to architectural design, a prototype implemented and presents results obtained and illustrates its operation in an example implementation. Finally, there are a discuss the results and future prospects for the system.
O aumento da demanda pela modalidade de Educação a Distância e o surgimento e aperfeiçoamento dos Ambientes Virtuais de Aprendizagem (AVA) apresentam-se num cenário em que se pretende minimizar distâncias e dinamizar estratégias de monitoramento dos alunos. No entanto, os AVA atuais não dispõem de recursos que identifiquem os usuários durante o desenvolvimento das suas atividades. Esta dissertação apresenta um Sistema de Identificação Automática de Faces (SIAF-EAD) para uso em AVA que possui uma arquitetura cliente-servidor e combina estágios de processamento de reconhecimento facial com módulos de um mecanismo que monitora usuários remotos através de uma webcam. Para realizar o reconhecimento facial foi utilizada uma abordagem de seleção de coeficientes da Transformada Discreta do Cosseno (DCT - Discrete Cosine Transform) que obteve 97,12% de taxa de acertos em um banco de imagens coletadas sem normalização das faces. Este trabalho descreve, além do projeto arquitetural, um protótipo implementado e apresenta resultados obtidos e ilustra seu funcionamento em um exemplo de execução. Por fim, são discutidos os resultados obtidos e perspectivas futuras para o sistema.

Traitant des données très sensibles, les systèmes de gestion d'identité doivent fournir une protection adéquate de la confidentialité. En utilisant le calcul multipartite (MPC), le chiffrement homomorphe (HE) et le chiffrement fonctionnel (FE), cette thèse aborde la conception et la mise en œuvre de systèmes biométriques préservant la confidentialité pour de multiples scénarios. Nous améliorons les travaux existants dans le domaine, en équilibrant la précision et la performance avec les garanties de sécurité. Nous allons au-delà des adversaires semi-honnêtes pour garantir la correction face aux adversaires malveillants. Enfin, nous abordons la question de la fuite des données biométriques lors de la révélation du résultat, un problème de confidentialité souvent négligé dans la littérature. Les principales contributions de cette thèse sont : - Une nouvelle solution d'identification de visage construite sur la FE pour produits scalaires atténuant la fuite d'entrée. - Un nouveau protocole de calcul à deux parties, Funshade, pour préserver la confidentialité des opérations biométriques de calcul de distance avec seuil. - Une méthode innovante d'identification biométrique préservant la confidentialité, basée sur la notion de test de groupe appelée Grote. - Un nouveau protocole de décryptage distribué avec masquage collaboratif traitant la fuite d'entrée, appelé Colmade. - Un protocole de calcul tripartite à majorité honnête, Banners, pour réaliser l'inférence malicieusement sécurisée de réseaux neuronaux binarisés. - Une bibliothèque Python HE nommée Pyfhel, offrant une abstraction de haut niveau et des fonctionnalités de bas niveau, avec des applications dans l'enseignement
Dealing with highly sensitive data, identity management systems must provide adequate privacy protection as they leverage biometrics technology. Wielding Multi-Party Computation (MPC), Homomorphic Encryption (HE) and Functional Encryption (FE), this thesis tackles the design and implementation of practical privacy-preserving biometric systems, from the feature extraction to the matching with enrolled users. This work is consecrated to the design of secure biometric solutions for multiple scenarios, putting special care to balance accuracy and performance with the security guarantees, while improving upon existing works in the domain. We go beyond privacy preservation against semi-honest adversaries by also ensuring correctness facing malicious adversaries. Lastly, we address the leakage of biometric data when revealing the output, a privacy concern often overlooked in the literature. The main contributions of this thesis are: • A new face identification solution built on FE-based private inner product matching mitigating input leakage. • A novel efficient two-party computation protocol, Funshade, to preserve the privacy of biometric thresholded distance metric operations. • An innovative method to perform privacy-preserving biometric identification based on the notion of group testing named Grote. • A new distributed decryption protocol with collaborative masking addressing input leakage, dubbed Colmade. • An honest majority three-party computation protocol, Banners, to perform maliciously secure inference of Binarized Neural Networks. • A HE Python library named Pyfhel, offering a high-level abstraction and low-level functionalities, with applications in teaching

Mestrado de dupla diplomação com a UTFPR - Universidade Tecnológica Federal do Paraná
Os sistemas de controle de presenças que realizam a autenticação através de faces carecem de detectores de fraudes para que sejam mais confiáveis. Um sistema capaz de executar essa tarefa automaticamente e corretamente vem trazer uma série de vantagens práticas no domínio da autenticação biométrica. Para atender esta carência, um detector de face falsa é desenvolvido e serve como um pré-passo antes do reconhecimento facial. A abordagem proposta para detecção de face falsa é utilizar câmera infravermelha do espectro NIR e machine learning, referida de deep learning. Neste trabalho foi criado uma base de dados de imagens de faces falsas e reais com auxílio de uma câmera com luz infravermelha NIR. A partir das imagens, foram gerados três datasets para implementação dos modelos de machine learning: Árvore de Decisão, Random Forest, KNN, SVM e MLP. Para a construção do protótipo de reconhecimento facial com detector de face falsa foi utilizado a linguagem Python de programação, as bibliotecas de programação: OpenFace, Scikit- Learn, OpenCV e Flask. A partir destas ferramentas e modelos treinados foi possível ter uma acurácia de 97.50% para detecção de faces falsas e faces reais com o classificador SVM. Para o reconhecimento facial foi definido uma limiar (de 0 a 1) confiável de 0.6 para sistemas que utilizam autenticação no formato 1 para N e limiar 0.2 para formato 1 para 1. Pretende-se que no futuro, o protótipo proposto seja ensaiado numa rede de terminais de marcação de presenças no IPB.
Presence control systems that use perform face authentication need fraud detectors more reliable. A system to able to detect this task automatically and correctly brings a number of practical advantages in the field of biometric authentication. For this problem, an anti-spoofing is developed and serves as a pre-step before face recognition. The proposed approach for false face detection is to use NIR infrared camera and machine learning with deep learning. In this dissertation, it was created a database of fake and real face images with an infrared camera. From the images, three datasets were created to implement the machine learning models: Decision Tree, Random Forest, KNN, SVM and MLP. For the construction of the face recognition prototype with anti-spoofing, the Python programming language, the OpenFace, Scikit-Learn, OpenCV and Flask programming libraries were used. From these trained tools and models it was possible to have an accuracy of 97.50% for detection of false faces and real faces with the SVM classifier. For face recognition, a reliable threshold (from 0 to 1) of 0.6 for systems using 1 to N format authentication and 0.25 to 1 to 1 format threshold is set. It is intended that the proposed prototype be tested on a network of attendance at IPB.