Gonzalo Bailador

Application of the computational theory of perceptions to human gait pattern recognition

This paper aims to contribute to the field of human gait pattern recognition by providing a solution based on the computational theory of perceptions. Our model differs significantly from others, e.g., based on machine learning techniques, because we use a linguistic model to represent the subjective designers perceptions of the human gait process. This model is easily understood and provides good results. We include a practical demonstration with an equal error rate of 3%.

Analysis of pattern recognition techniques for in-air signature biometrics

As a result of advances in mobile technology, new services which benefit from the ubiquity of these devices are appearing. Some of these services require the identification of the subject since they may access private user information. In this paper, we propose to identify each user by drawing his/her handwritten signature in the air (in-air signature). In order to assess the feasibility of an in-air signature as a biometric feature, we have analysed the performance of several well-known pattern recognition techniques-Hidden Markov Models, Bayes classifiers and dynamic time warping-to cope with this problem. Each technique has been tested in the identification of the signatures of 96 individuals. Furthermore, the robustness of each method against spoofing attacks has also been analysed using six impostors who attempted to emulate every signature. The best results in both experiments have been reached by using a technique based on dynamic time warping which carries out the recognition by calculating distances to an average template extracted from several training instances. Finally, a permanence analysis has been carried out in order to assess the stability of in-air signature over time.

Pattern recognition using temporal fuzzy automata

In this paper, we propose a syntactic pattern recognition approach based on fuzzy automata, which can cope with the variability of patterns by defining imprecise models. This approach is called temporal fuzzy automata as it allows the inclusion of time restrictions to model the duration of the different states. The concept of fuzzy state makes it possible to handle ambiguity as the automaton can be in several states at the same time. Another advantage of our approach is the capability to synchronize with the signal, which allows us to avoid the segmentation stage before the recognition process. Furthermore, a learning method based on dynamic time warping is provided that makes it possible to automatically generate models. Finally, to demonstrate the performance and robustness of this approach, we have applied it to the recognition of hand gestures without any kind of signal preprocessing.

Authentication in mobile devices through hand gesture recognition

This article proposes an innovative biometric technique based on the idea of authenticating a person on a mobile device by gesture recognition. To accomplish this aim, a user is prompted to be recognized by a gesture he/she performs moving his/her hand while holding a mobile device with an accelerometer embedded. As users are not able to repeat a gesture exactly in the air, an algorithm based on sequence alignment is developed to correct slight differences between repetitions of the same gesture. The robustness of this biometric technique has been studied within 2 different tests analyzing a database of 100 users with real falsifications. Equal Error Rates of 2.01 and 4.82% have been obtained in a zero-effort and an active impostor attack, respectively. A permanence evaluation is also presented from the analysis of the repetition of the gestures of 25 users in 10 sessions over a month. Furthermore, two different gesture databases have been developed: one made up of 100 genuine identifying 3-D hand gestures and 3 impostors trying to falsify each of them and another with 25 volunteers repeating their identifying 3-D hand gesture in 10 sessions over a month. These databases are the most extensive in published studies, to the best of our knowledge.

Analysis of pattern recognition and dimensionality reduction techniques for odor biometrics

In this paper, we analyze the performance of several well-known pattern recognition and dimensionality reduction techniques when applied to mass-spectrometry data for odor biometric identification. Motivated by the successful results of previous works capturing the odor from other parts of the body, this work attempts to evaluate the feasibility of identifying people by the odor emanated from the hands. By formulating this task according to a machine learning scheme, the problem is identified with a small-sample-size supervised classification problem in which the input data is formed by mass spectrograms from the hand odor of 13 subjects captured in different sessions. The high dimensionality of the data makes it necessary to apply feature selection and extraction techniques together with a simple classifier in order to improve the generalization capabilities of the model. Our experimental results achieve recognition rates over 85% which reveals that there exists discriminatory information in the hand odor and points at body odor as a promising biometric identifier.

Time series distances measures to analyze in-air signatures to authenticate users on mobile phones

Improving the security of mobile phones is one of the crucial points required to assure the personal information and the operations that can be performed from them. This article presents an authentication procedure consisting of verifying the identity of people by making a signature in the air while holding the mobile phone. Different temporal distance algorithms have been proposed and evaluated through a database of 50 people making their signatures in the air and 6 people trying to forge each of them by studying their records. Approaches based on DTW have obtained better EER results than those based on LCS (2.80% against 3.34%). Besides, different signal normalization methods have been evaluated not finding any with better EER results that when no normalization has carried out.

Low computational cost multilayer graph-based segmentation algorithms for hand recognition on mobile phones

Unconstrained and contact-free hand recognition problem with mobile devices is not solved yet because these systems have to deal with hard problems like different backgrounds and illumination. Algorithms to perform an image segmentation in order to create regions in the image with the same semantic meaning are a work in progress. Graph theory has been used successfully in order to reach a good image segmentation in many fields but these algorithms are computational demanding (time and memory) making it very difficult to use on mobile platforms. New algorithms to perform image segmentation are needed in order to adapt biometric technologies to mobile devices. This paper presents a segmentation algorithm based on multilayer graphs. We compared our results with other known segmentation algorithms (NCuts and KMeans) by using a synthetic database with over 400000 images. Our results show that the optimized implementation of the proposed algorithm makes this a powerful tool with high accuracy and low computational cost, improving the accuracy and the execution time of the two other algorithms.

Secure access control by means of human stress detection

This paper proposes a stress detection system based on fuzzy logic and the physiological signals heart rate and galvanic skin response. The main contribution of this method relies on the creation of a stress template, collecting the behaviour of previous signals under situations with a different level of stress in each individual. The creation of this template provides an accuracy of 99.5% in stress detection, improving the results obtained by current pattern recognition techniques like GMM, k-NN, SVM or Fisher Linear Discriminant. In addition, this system can be embedded in security systems to detect critical situations in accesses as cross-border control. Furthermore, its applications can be extended to other fields as vehicle driver state-of-mind management, medicine or sport training.

gb2sμMOD: A MUltiMODal biometric video database using visible and IR light

Abstract In spite of recent efforts in gathering multimodal databases containing a big number of traits, a huge amount of users and covering multiple realistic scenarios, there is still a lack of touch-less realistic samples, video recordings for some traits and the use of infrared cameras which allows, among others, to avoid lighting influence and test recently appeared biometric techniques such as hand vein recognition. For this reason, a new realistic multimodal database composed of 8,160 hand, iris and face videos has been captured. To this end, a total of 60 contributors have participated in three separated acquisition sessions in which three different cameras have been used, covering different ranges of the light spectrum: visible light and two different infrared wavelengths. To simulate real-world working conditions, the database has been recorded in an indoor environment with different lightings and backgrounds. In addition, due to the relevance of performing evaluation experiments in such a way that a reliable comparison of the results can be accomplished, an evaluation protocol is provided at the end of this paper. Moreover, performance results are provided for several biometric traits in mono- and multi- modalities that can be used as a baseline.

A robustness verification system for mobile phone authentication based on gestures using Linear Discriminant Analysis

This article evaluates an authentication technique for mobiles based on gestures. Users create a remindful identifying gesture to be considered as their in-air signature. This work analyzes a database of 120 gestures of different vulnerability, obtaining an Equal Error Rate (EER) of 9.19% when robustness of gestures is not verified. Most of the errors in this EER come from very simple and easily forgeable gestures that should be discarded at enrollment phase. Therefore, an in-air signature robustness verification system using Linear Discriminant Analysis is proposed to infer automatically whether the gesture is secure or not. Different configurations have been tested obtaining a lowest EER of 4.01% when 45.02% of gestures were discarded, and an optimal compromise of EER of 4.82% when 19.19% of gestures were automatically rejected.