This thesis describes research into effective voice biometrics (speaker recognition) under mismatched noise conditions. Over the last two decades, this class of biometrics has been the subject of considerable research due to its various applications in such areas as telephone banking, remote access control and surveillance. One of the main challenges associated with the deployment of voice biometrics in practice is that of undesired variations in speech characteristics caused by environmental noise. Such variations can in turn lead to a mismatch between the corresponding test and reference material from the same speaker. This is found to adversely affect the performance of speaker recognition in terms of accuracy. To address the above problem, a novel approach is introduced and investigated. The proposed method is based on minimising the noise mismatch between reference speaker models and the given test utterance, and involves a new form of Test-Normalisation (T-Norm) for further enhancing matching scores under the aforementioned adverse operating conditions. Through experimental investigations, based on the two main classes of speaker recognition (i.e. verification/ open-set identification), it is shown that the proposed approach can significantly improve the performance accuracy under mismatched noise conditions. In order to further improve the recognition accuracy in severe mismatch conditions, an approach to enhancing the above stated method is proposed. This, which involves providing a closer adjustment of the reference speaker models to the noise condition in the test utterance, is shown to considerably increase the accuracy in extreme cases of noisy test data. Moreover, to tackle the computational burden associated with the use of the enhanced approach with open-set identification, an efficient algorithm for its realisation in this context is introduced and evaluated. The thesis presents a detailed description of the research undertaken, describes the experimental investigations and provides a thorough analysis of the outcomes.

This thesis examines the main implications, as well as future possibilities and challenges of a high-tech ID-system for refugees in Uganda. The implemented system captures biometric information and targets the UN sustainable development goal 16.9 legal identity for all. Through examining the involved parties’ experiences, perceptions and attitudes towards the digital ID system, the study contextualizes a global high-tech system in local rural areas. It moreover highlights the importance of social and geographical factors. The thesis is a result of a qualitative field study in Uganda where governmental and NGO representatives, working in the refugee settlements, were interviewed. The findings show that the IDs have improved the Ugandan refugee response and function as a base for delivering and receiving assistance in terms of food and cash distribution as well as access to SIM-cards. The IDs also give the providers more accurate data about the refugees, enabling population tracking for protection. The data is however sensitive and vulnerable to bias and misusage. Lastly, the provided IDs are important for an increased inclusion of refugees in Uganda aligned with the global goal.

Orientador: Helenice de Oliveira Florentino Silva / Banca: Diana Rodrigues de Pina / Banca: Maria do Socorro Nogueira Rangel / Resumo: Um planejamento para radioterapia é considerado ótimo quando todos os parâmetros envolvidos sejam eles físicos ou biológicos, foram investigados e adequados individualmente para cada paciente. Neste tipo de planejamento, a grande preocupação é com a irradiação do tumor com o mínimo dano possível aos tecidos saudáveis da região irradiada, principalmente aos órgãos de riscos. O planejamento ótimo para radioterapia pode ser auxiliado pela Programação Linear e existe uma ampla literatura abordando este assunto, mas, a maioria das formulações matemáticas publicadas não contemplam um cenário do ponto de vista de aplicações práticas, pois não incorporam determinados fatores que são de extrema importância para a construção de um planejamento real, como exemplos a atenuação do feixe de radiação e a beterogeneidade na composição dos tecidos irradiados. Assim, este trabalho apresenta uma metodologia para correção de heterogeneidade na composição dos diferentes tipos de tecidos irradiados baseado nas proporções entre seus diferentes coeficientes de atenuação linear. Esta metodologia tem como objetivo tornar as simulações de planejamentos otimizados mais próximos dos planejamentos reais e desta forma, possibilitar um estudo mais amplo e confiável, faznedo com que modelos de programação linear possam ser utilizados como ferramentas auxiliares na criação de planejamentos reais para radioterapia / Abstract: Planning for radiotherapy is considered optimal when all the parameters involved, physical or biological, have been investigated and are appropriate for each patient. In this type of planning, the major concern is with the irradiation of the tumor with the minimum possible damage to healthly tissues of the irradiated region, especially the organs at risk. The optimal planning for radiation therapy can be aided by Linear Programming and there is a wide literature addressing this subject. However, most published mathematical formulations do not contemplate a scenario in terms of practical applications. They do not incorporate certain factors that are extremely important for building a real planning, as examples there are attenuation of the radiation beam and the heterogeneous composition of the irradiated tissue. This work presents a methodology for correction of heterogeneity in the composition of different types of tissuers irradiated based on the proportions among their different linear attenuation coefficient. This methodology aims to make the simulations of optimized planning closer to the real planning and thus enable a more comprehensive and reliable, allowing the use of linear programming models as aids in the creation of real planning for radiotherapy / Mestre

O Brasil é um dos maiores produtores e exportadores de carne bovina do planeta e com a expectativa de ser responsável por 45% do consumo mundial sendo que a maior parte de seu consumo ainda é local. Para se manter nesta posição e expandir suas vendas tanto no mercado interno como externo é importante que se garanta a qualidade do produto. Esta qualidade só é conseguida quando se consegue gerenciar todo o processo da cadeia produtiva, de forma a permitir o registro de todos os dados do animal na cadeia de produção. Tanto o governo através do SISBOV como grandes distribuidores possuem sistemas de gerenciamento que através de técnicas de rastreabilidade permitem ter este controle. A identificação do animal é ponto chave para a rastreabilidade que hoje é feita através de bottons, transponders, brincos entre outros. Todos estes métodos são invasivos e suscetíveis a perdas e adulterações. Esta dissertação mostra a viabilidade de inserir em sistemas de rastreabilidade existentes ou não a inclusão de identificação biométrica e usa como exemplo o focinho nasal do bovino. Para tanto desenvolveu-se programas para entrada de informações através de um celular com sistema operacional Android que em conjunto com programas desenvolvidos para rodarem na WEB pudessem cadastrar e confirmar a identidade do animal. Os testes mostraram a capacidade do aplicativo Android em localizar o espelho nasal e coletar o mesmo. Com os dados coletados foi possível armazenar as informações ou confirmar a identidade do animal por meio dos serviços do servidor. Mostrou-se desta forma viável a utilização deste tipo de identificação em sistemas de gerenciamento novos ou já existentes. / Brazil is one of the largest producers and exporters of beef in the world and with the expectation of being responsible for 45% of global consumption and the largest part of consumption is still locall. To stay in this position and to expand its sales both in domestic and foreign markets is important to ensure product quality. This quality is achieved only when you can manage the entire process of the production chain in order to allow the registration of all animal data in the production chain. The government through SISBOV and large distributors have been working with management systems through traceability techniques. Animal identification is key to the traceability and nowadays is made via bottoms, transponders, earrings among others. All these methods are invasive and susceptible to loss and tampering. This work shows the feasibility of entering into existing traceability systems or not the inclusion of biometric identification and uses as an example the nasal muzzle beef. For both developed programs to input information through a mobile phone with Android operating system in conjunction with programs designed to be web based solution could register and confirm the identity of the animal. All results was able to shown the system performance by showing that it was possible to store the information or confirm the identity of the animal through the server services. The methodology proposed could be useful in commercial applications focusing in bovine traceability.

This thesis is dedicated to extracting expression robust features for 3D face recognition. The use of 3D imaging enables the extraction of discriminative features that can significantly improve the recognition performance due to the availability of facial surface information such as depth, surface normals and curvature. Expression robust analysis using information from both depth and surface normals is investigated by dividing the main facial region into patches of different scales. The nasal region and adjoining parts of the cheeks are utilized as they are more consistent over different expressions and are hard to deliberately occlude. In addition, in comparison with other parts of the face, these regions have a high potential to produce discriminative features for recognition and overcome pose variations. An overview and classification methodology of the widely used 3D face databases are first introduced to provide an appropriate reference for 3D face database selection. Using the FRGC and Bosphorus databases, a low complexity pattern rejector for expression robust 3D face recognition is proposed by matching curves on the nasal and its environs, which results in a low-dimension feature set of only 60 points. To extract discriminative features more locally, a novel multi-scale and multi-component local shape descriptor is further proposed, which achieves more competitive performances under the identification and verification scenarios. In contrast with many of the existing work on 3D face recognition that consider captures obtained with laser scanners or structured light, this thesis also investigates applications to reconstructed 3D captures from lower cost photometric stereo imaging systems that have applications in real-world situations. To this end, the performance of the expression robust face recognition algorithms developed for captures from laser scanners are further evaluated on the Photoface database, which contains naturalistic expression variations. To improve the recognition performance of all types of 3D captures, a universal landmarking algorithm is proposed that makes uses of different components of the surface normals. Using facial profile signatures and thresholded surface normal maps, facial roll and yaw rotations are calibrated and five main landmarks are robustly detected on the well-aligned 3D nasal region. The landmarking results show that the detected landmarks demonstrate high within-class consistency and can achieve good recognition performances under different expressions. This is also the first landmarking work specifically developed for the reconstructed 3D captures from photometric stereo imaging systems.

Biometrics are a convenient alternative to traditional forms of access control such as passwords and pass-cards since they rely solely on user-specific traits. Unlike alphanumeric passwords, biometrics cannot be given or told to another person, and unlike pass-cards, are always “on-hand.” Perhaps the most well-known biometrics with these properties are: face, speech, iris, and gait. This dissertation proposes a new biometric modality: gestures. A gesture is a short body motion that contains static anatomical information and changing behavioral (dynamic) information. This work considers both full-body gestures such as a large wave of the arms, and hand gestures such as a subtle curl of the fingers and palm. For access control, a specific gesture can be selected as a “password” and used for identification and authentication of a user. If this particular motion were somehow compromised, a user could readily select a new motion as a “password,” effectively changing and renewing the behavioral aspect of the biometric. This thesis describes a novel framework for acquiring, representing, and evaluating gesture passwords for the purpose of general access control. The framework uses depth sensors, such as the Kinect, to record gesture information from which depth maps or pose features are estimated. First, various distance measures, such as the log-euclidean distance between feature covariance matrices and distances based on feature sequence alignment via dynamic time warping, are used to compare two gestures, and train a classifier to either authenticate or identify a user. In authentication, this framework yields an equal error rate on the order of 1-2% for body and hand gestures in non-adversarial scenarios. Next, through a novel decomposition of gestures into posture, build, and dynamic components, the relative importance of each component is studied. The dynamic portion of a gesture is shown to have the largest impact on biometric performance with its removal causing a significant increase in error. In addition, the effects of two types of threats are investigated: one due to self-induced degradations (personal effects and the passage of time) and the other due to spoof attacks. For body gestures, both spoof attacks (with only the dynamic component) and self-induced degradations increase the equal error rate as expected. Further, the benefits of adding additional sensor viewpoints to this modality are empirically evaluated. Finally, a novel framework that leverages deep convolutional neural networks for learning a user-specific “style” representation from a set of known gestures is proposed and compared to a similar representation for gesture recognition. This deep convolutional neural network yields significantly improved performance over prior methods. A byproduct of this work is the creation and release of multiple publicly available, user-centric (as opposed to gesture-centric) datasets based on both body and hand gestures.

O tema geral desta pesquisa são as tecnologias que permitem o controle de acesso, vigilância, monitoramento e identificação de pessoas, e que se aliam a construção de bancos de dados e perfis sobre a população. Neste amplo universo, a tecnologia biométrica para identificação foi focalizada a partir de um estudo de caso sobre o novo documento biométrico de identidade brasileiro: o Registro de Identidade Civil. Retomando o conceito de dispositivo em Michel Foucault, buscou-se trazer a tona os discursos, as instituições, as leis, o debate legal, as medidas, decisões, e enunciados científicos que configuram o funcionamento do poder na atualidade. No âmbito das ciências, a biometria hoje distancia-se da antropometria e das formas de identificação do século XIX, vinculando-se a um exercício do poder que não é mais aquele para disciplinar os corpos (Michel Foucault), mas para gerir os fluxos de dados, um corpo de dados. As novas tecnologias focalizadas apontam para um exercício do poder mais próximo do que Gilles Deleuze chamou de sociedades de controle. / This research focus on technologies that are enabled to access control, surveillance, monitoring and identification of persons, connected with databases and profiles construction on the population. In this vast universe, the biometric technology for identification has been focused from a case study on the new biometric identity document Brazil: the Civil Identity Register. Based on the Foucaults concept of apparatus, this reasearch aimed to bring out the discourses, institutions, laws, the legal debate, measures, decisions, and scientific statements that configure the operation of power today. Within the sciences, biometrics today area distanced itself from anthropometry and forms of identification of the nineteenth century. This is related with an exercise of power that is no longer that to discipline their bodies (Michel Foucault), but to manage the data flows, or a \"body of data. New technologies are related with an exercise of power closer to what Gilles Deleuze called societies of control.

abstract: Information forensics and security have come a long way in just a few years thanks to the recent advances in biometric recognition. The main challenge remains a proper design of a biometric modality that can be resilient to unconstrained conditions, such as quality distortions. This work presents a solution to face and ear recognition under unconstrained visual variations, with a main focus on recognition in the presence of blur, occlusion and additive noise distortions. First, the dissertation addresses the problem of scene variations in the presence of blur, occlusion and additive noise distortions resulting from capture, processing and transmission. Despite their excellent performance, ’deep’ methods are susceptible to visual distortions, which significantly reduce their performance. Sparse representations, on the other hand, have shown huge potential capabilities in handling problems, such as occlusion and corruption. In this work, an augmented SRC (ASRC) framework is presented to improve the performance of the Spare Representation Classifier (SRC) in the presence of blur, additive noise and block occlusion, while preserving its robustness to scene dependent variations. Different feature types are considered in the performance evaluation including image raw pixels, HoG and deep learning VGG-Face. The proposed ASRC framework is shown to outperform the conventional SRC in terms of recognition accuracy, in addition to other existing sparse-based methods and blur invariant methods at medium to high levels of distortion, when particularly used with discriminative features. In order to assess the quality of features in improving both the sparsity of the representation and the classification accuracy, a feature sparse coding and classification index (FSCCI) is proposed and used for feature ranking and selection within both the SRC and ASRC frameworks. The second part of the dissertation presents a method for unconstrained ear recognition using deep learning features. The unconstrained ear recognition is performed using transfer learning with deep neural networks (DNNs) as a feature extractor followed by a shallow classifier. Data augmentation is used to improve the recognition performance by augmenting the training dataset with image transformations. The recognition performance of the feature extraction models is compared with an ensemble of fine-tuned networks. The results show that, in the case where long training time is not desirable or a large amount of data is not available, the features from pre-trained DNNs can be used with a shallow classifier to give a comparable recognition accuracy to the fine-tuned networks. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2018

<p>Different security issues are a top subject around the world, especially since the terror threats seem to intensify. In the same time, the transport industry suffer from problems with smuggling and theft of valuable goods. One way to increase the security might be to have a verification system installed in commercial trucks, in order to assure that the driver is the proper one.</p><p>This thesis has two purposes. One is to find appropriate methods for driver verification and build a prototype of a verification system which can be used for testing and further development. The other is to study how truck drivers perceive such a system and how their conception goes along with the growing demand for higher security. The present work is the result of a cooperation between an engineer and a cognitive scientist. The thesis focuses on the transport industry and was performed for Volvo Technology Corporation (VTEC), Gothenburg, Sweden.</p><p>Eleven available verification methods were studied. To enable a well-based selection of methods to implement in the prototype, inquiries and interviews with truck drivers and haulage contractors were carried out to complement the theoretical study. </p><p>One regular and three biometric verification methods were chosen for the test; fingerprint verification, face recognition, voice recognition and PIN verification. These methods were put together to a prototype system that was implemented in a truck simulator. A graphical user interface was developed in order to make the system user friendly. The prototype system was tested by 18 truck drivers. They were thoroughly interviewed before and after the test in order to retrieve their background, expectations and opinions as well as their perceptions and experiences of the test. </p><p>Most of the test participants were positive to the prototype system. Even though they did not feel a need for it today they believed it to “be the future”. However, some participants felt uncomfortable with the system since they felt controlled by it. It became clear how important it is to have a system that respect the users’ privacy and to assure that the users are well informed about how the system is used. Some of the technology used for the verification system requires more development to fit in the automotive context, but it is considered to be possible to achieve a secure and robust system.</p>

The technology exists for the migration of healthcare data from its archaic paper-based system to an electronic one and once in digital form, to be transported anywhere in the world in a matter of seconds. The advent of universally accessible healthcare data benefits all participants, but one of the outstanding problems that must be addressed is how to uniquely identify and link a patient to his or her specific medical data. To date, a few solutions to this problem have been proposed that are limited in their effectiveness. We propose the use of biometric technology within our FIRD framework in solving the unique association of a patient to his or her medical data distinctively. This would allow a patient to have real time access to all of his or her recorded healthcare information electronically whenever it is necessary, securely with minimal effort, greater effectiveness, and ease.