In this thesis, a new algorithm has been proposed to segment the foreground of the fingerprint from the image under consideration. The algorithm uses three features, mean, variance and coherence. Based on these features, a rule system is built to help the algorithm to efficiently segment the image. In addition, the proposed algorithm combine split and merge with modified Otsu. Both enhancements techniques such as Gaussian filter and histogram equalization are applied to enhance and improve the quality of the image. Finally, a post processing technique is implemented to counter the undesirable effect in the segmented image. Fingerprint recognition system is one of the oldest recognition systems in biometrics techniques. Everyone have a unique and unchangeable fingerprint. Based on this uniqueness and distinctness, fingerprint identification has been used in many applications for a long period. A fingerprint image is a pattern which consists of two regions, foreground and background. The foreground contains all important information needed in the automatic fingerprint recognition systems. However, the background is a noisy region that contributes to the extraction of false minutiae in the system. To avoid the extraction of false minutiae, there are many steps which should be followed such as preprocessing and enhancement. One of these steps is the transformation of the fingerprint image from gray-scale image to black and white image. This transformation is called segmentation or binarization. The aim for fingerprint segmentation is to separate the foreground from the background. Due to the nature of fingerprint image, the segmentation becomes an important and challenging task. The proposed algorithm is applied on FVC2000 database. Manual examinations from human experts show that the proposed algorithm provides an efficient segmentation results. These improved results are demonstrating in diverse experiments.

Made available in DSpace on 2015-04-22T22:00:50Z (GMT). No. of bitstreams: 1 Thiago Falcao.pdf: 1370921 bytes, checksum: ec7b9ab219f2028eded75407403140be (MD5) Previous issue date: 2014-01-13 / This work proposes the new classifier for face recognition, novelty classifier, which is based on novelty filter proposed by Kohonen. In order to evaluate the new classifier performance, it is performed a comparison with nearest neighboard classifier, which uses the Euclidian distance as distance metric. ORL face database was chosen to be used in this comparison. There was not any pre-processing (photometric or geometric) on face images. It was used the following feature extraction methods: PCA, 2DPCA and (2D)2PCA. Some results in identification mode are exposed through rank 1 recognition rate and CMC curves. In verification mode, the results were presented by Correct Acceptance Rate (CAR), Equivalent Error Rate (EER), ROC curves and Area under the ROC curve (AUC). Results shown that the proposed classifier performs better than others previously published, when the 10-fold Cross Validation method is employed as a test strategy. Recognition rate of 100% is achieved with this test methodology. / Este trabalho propõe a utilização do classificador de novidades para reconhecimento de faces, o qual é baseado no filtro de novidades, proposto por Kohonen. Para avaliar o desempenho do novo classificador é feita uma comparação com o classificador do vizinho mais próximo, usando a métrica da distância euclidiana. A base de dados utilizada para essa comparação foi a base ORL. A informação da face é extraída utilizando os métodos PCA, 2DPCA e (2D)2PCA, sem usar qualquer tipo de pré-processamento (fotométrico ou geométrico). Os seguintes resultados são apresentados no modo de identificação: taxa de reconhecimento rank 1 e as curvas CMC, no modo verificação: as taxas de correta aceitação (CAR), de erro equivalente (EER), as curvas ROC e área sob a curva ROC (AUC). Os resultados obtidos mostraram que o classificador proposto tem um desempenho melhor do que o desempenho do vizinho mais próximo e do que outros classificadores anteriormente publicados usando a mesma base, quando a estratégia de validação cruzada 10-fold é usada, com essa estratégia a taxa de reconhecimento obtida foi de 100%

Baigiamajame magistro darbe nagrinėjamas genetinių algoritmų taikymas biometrijoje. Išnagrinėta plačiausiai naudojama biometrinė informacija, aprašytos labiausiai paplitusios biometrinės sistemos, genetiniai algoritmai bei jų pritaikymas biometrinių sistemų optimizavimui. Baigiamajame darbe pasiūlytas naujas rankos atpažinimo metodas. Šiam metodui pritaikyti genetiniai algoritmai. Darbą sudaro 7 dalys: įvadas, biometrija, genetiniai algoritmai, genetinių algoritmų taikymas biometrinėse sistemose, genetinių algoritmų taikymas rankos atpažinimui, išvados ir literatūra. Darbo apimtis – 51 p. teksto be priedų, 30 pav., 4 lent., 32 bibliografiniai šaltiniai. / In the graduation thesis to receive the master‘s degree the application of genetic algorithms in biometrics is analysed. The most widely used biometric information have been examined, the most common biometric systems, genetic algorithms and their customization in biometric systems optimization have been described. A new method is proposed for hand recognition. Genetic algorithms have been customized for this method. Structure: introduction, biometry, genetic algorithms, application of genetic algorithms in biometric systems, application of genetic algorithms for hand recognition, the conclusions and bibliography. Thesis consist of: 51 p. text without appendixes, 30 pictures, 4 tables, 32 bibliographical entries.

Currently living in the digital age, so the manipulation of data and images is often all day. Due to the problem of space for storage of pictures and time of transmission, many compression techniques had been developed, and a great challenge is to make these techniques to bring good results in terms of compression rate, picture quality and processing time. The Fractal Compression technique developed by Fisher, was described, implemented and tested in this work and it brought great results, and considerable improvement in terms of execution time, which was rather low. Another area that has been emphasizing is the use of biometric techniques to the people recognition. A very used technique is the iris recognition that has shown enough reliability. Thus, connecting the two technologies brings great benefits. In this work, images of iris were compressed by the method implemented here and were made simulations of the technique iris recognition developed by Libor Maseck. The results show that it is possible to compress fractally the images without damage the recognition system. Comparisons were made and was possible realize that even with changes in pixels of images, the system remains very reliable, bringing benefits to storage space. / Atualmente vive-se na era digital, por isso a manipulação de dados e imagens é freqüente todos os dias. Devido ao problema de espaço para armazenamento dessas imagens e tempo de transmissão, foram desenvolvidas várias técnicas de compressão, e um grande desafio é fazer com que essas técnicas tragam bons resultados em termos de taxa de compressão, qualidade da imagem e tempo de processamento. A técnica de compressão Fractal desenvolvida por Fisher, foi descrita, implementada e testada neste trabalho e trouxe ótimos resultados, e melhoria considerável em termos de tempo de execução, que foi bastante reduzido. Outra área que vem se destacando é o uso de técnicas biométricas para reconhecimento de pessoas. Uma técnica muito usada é o reconhecimento de íris que tem mostrado bastante contabilidade. Assim, aliar as duas tecnologias traz grandes benefícios. No presente trabalho, imagens de íris foram comprimidas pelo método aqui implementado e foram realizadas simulações da técnica de reconhecimento de íris desenvolvida por Maseck. Os resultados mostram que é possível comprimir fractalmente as imagens sem prejudicar o sistema de reconhecimento. Comparações foram realizadas e foi possível perceber que mesmo havendo mudanças nos pixels das imagens, o sistema permanece bastante confiavel, trazendo vantagens em espaço de armazenamento. / Mestre em Ciências

The main topic of the thesis is creation of formulas for the amount of information entropy in biometric characteristics of iris and retina. This field of science in biometrics named above is unstudied yet, so the thesis tries to initiate research in this direction. The thesis also discusses the historical context of security and identification fields according to biometric characteristics of a human being with an overlap for potential usage in biometrics of iris and retina. The Daugman’s algorithm for converting iris image into the binary code which can be used as a cryptographic key is discussed in detail. An application implementing this conversion is also a part of the thesis.

The aim of this thesis is a design and realization of the core of multimodal biometric system. First part of the thesis sumarizes contemporary knowledge about biometric systems and about combination of their outputs. Second part introduces concept and implementation of multimodal biometric system, which uses weighted score combination and user-specific weights.

This master‘s thesis deals with biometric fingerprint liveness detection. The theoretical part of the work describes fingerprint recognition biometric systems, fingerprint liveness detection issues and methods for fingerprint liveness detection. The practical part of the work describes proposed set of discriminant features and preprocessing of fingerprint image. Proposed approach using neural network to detect a liveness. The algorithm is tested on LivDet database comprising real and fake images acquired with tree sensors. Classification performance approximately 93% was obtained.

Voice recognition has become a more focused and researched field in the last century,and new techniques to identify speech has been introduced. A part of voice recognition isspeaker verification which is divided into Front-end and Back-end. The first componentis the front-end or feature extraction where techniques such as Mel-Frequency CepstrumCoefficients (MFCC) is used to extract the speaker specific features of a speech signal,MFCC is mostly used because it is based on the known variations of the humans ear’scritical frequency bandwidth. The second component is the back-end and handles thespeaker modeling. The back-end is based on the Gaussian Mixture Model (GMM) andGaussian Mixture Model-Universal Background Model (GMM-UBM) methods forenrollment and verification of the specific speaker. In addition, normalization techniquessuch as Cepstral Means Subtraction (CMS) and feature warping is also used forrobustness against noise and distortion. In this paper, we are going to build a speakerverification system and experiment with a variance in the amount of training data for thetrue speaker model, and to evaluate the system performance. And further investigate thearea of security in a speaker verification system then two methods are compared (GMMand GMM-UBM) to experiment on which is more secure depending on the amount oftraining data available.This research will therefore give a contribution to how much data is really necessary fora secure system where the False Positive is as close to zero as possible, how will theamount of training data affect the False Negative (FN), and how does this differ betweenGMM and GMM-UBM.The result shows that an increase in speaker specific training data will increase theperformance of the system. However, too much training data has been proven to beunnecessary because the performance of the system will eventually reach its highest point and in this case it was around 48 min of data, and the results also show that the GMMUBM model containing 48- to 60 minutes outperformed the GMM models.

This document deals with presentation of nowadays software and hardware methods used for fingerprint recognition with focus on liveness testing and thereafter it deals with description of my solution. In order to describe results obtained from study of technical literature, we discuss important terminology of biometric systems at first and further main principles of fingerprint sensors used in practice are shown. From overviewed methods of liveness detection we underline one method based on  perspiration (researched by BioSAL laboratory) and one spectroscopic method researched by Lumidigm Corporation. The study of liveness testing methods inspired me to creation of new type fingerprint sensor which has built-in livennes testing method based on two characteristic properties of living human tisue. In order to test this sensor, we discuss nowadays sensor deception method. It follows from their analysis, that newly designed sensor should be theoretically resistant to each of them.

The goal of the master thesis is the design and construction of a device for capturing video sequences of the cardiovascular system of the finger of a human hand and the subsequently design and implementation of a method of data extraction for its reconstruction into a 3D model.