Search

International audience; The next generation military environment requires a delay-tolerant network for sharing data and resources using an interoperable computerized, Command, Control, Communications, Intelligence, Surveillance and Reconnaissance (C4ISR) infrastructure. In this paper, we propose a new distributed SDN (Software-Defined Networks) architecture for tactical environments based on distributed cloudlets. The objective is to reduce the end-to-end delay of tactical traffic flow, and improve management capabilities, allowing flexible control and network resource allocation. The proposed SDN architecture is implemented over three layers: decentralized cloudlets layer where each cloudlet has its SDRN (Software-Defined Radio Networking) controller, decentralized MEC (Mobile Edge Computing) layer with an SDN controller for each MEC, and a centralized private cloud as a trusted third-part authority controlled by a centralized SDN controller. The experimental validations are done via relevant and realistic tactical scenarios based on strategic traffics loads, i.e., Tactical SMS (Short Message Service), UVs (Unmanned Vehicle) patrol deployment and high bite rate ISR (Intelligence, Surveillance, and Reconnaissance) video.

International audience; The accuracy of Visible Wavelength (VW) iris recognition systems is significantly affected by the quality of iris sample images. Using degraded VW iris sample images can significantly decrease the performance of iris recognition systems. There are not many iris quality enhancement methods in the research field. Therefore, it is interesting to investigate whether existing image quality enhancement methods can improve the performance ofbiometric systems for degraded VW iris images. In this paper, we apply nine image enhancement methods to a VW iris database, which contains high quality and degraded iris sample images. The experimental results show that some of selected enhancement methods can improve VW irisrecognition system performance.

International audience; La dynamique de frappe au clavier permet l'authentification et l'identification d'utilisateurs en analysant leur façon de taper au clavier, e.g. lorsqu'ils naviguent sur Internet. La majorité des études de l'état de l'art visent à améliorer les performances des systèmes de dynamique de frappe au clavier. Dans cet article, nous adressons le problème d'empêcher la capture de la dynamique de frappe afin de protéger la vie privée des utilisateurs. L'authentification/identification, ainsi que le profilage, peuvent être vu comme des attaques que nous limitons dans cette contribution. Des résultats expérimentaux, obtenus sur des bases de données significatives, montrent l'intérêt de la des méthodes proposées.

International audience; Keystroke Dynamics enables the authentication or identification of users by analyzing their way of typing, e.g.when browsing the Internet. Most studies in the state of the art focus on increasing Keystroke Dynamics Sys-tems performances. In this paper, we address the problematic of avoiding the biometric capture of keystrokedynamics in order to protect users’ privacy. Authentication/identification, profiling can be considered as at-tacks we limit in this contribution. Experimental results obtained on significant datasets show the benefits ofthe proposed approaches.

International audience; As a trending method for the authentication, biometrics tends to be integrated in various devices, and in particular in smartphones. If the evaluation is performed on operational device, the biometric sample and algorithm are not reachable by the assessors. So, these latter have to perform an evaluation on a system considered as a black box. This kind of evaluation implies numerous manual comparison. This paper proposes a methodology to perform an evaluation of biometric black boxes. In order to obtain this methodology, two experiments were performed in order to determine an optimized conduct. Nevertheless, these experiments were realized with small test population, in order to assess the methodology a full scale evaluation was performed. This paper describes the used methodology to perform evaluation on black boxes systems, and the results obtained on the systems under test.

International audience; The accuracy of face recognition systems is significantly affected by the quality of face sample images.There are many existing no-reference image quality metrics (IQMs) that are able to assess natural imagequality by taking into account similar image-based quality attributes. Previous study showed that IQMscan assess face sample quality according to the biometric system performance. In addition, re-trainingan IQM can improve its performance for face biometric images. However, only one database was used inthe previous study, and it contains only image-based distortions. In this paper, we propose to extend theprevious study by use multiple face database including FERET color face database and apply multiplesetups for the re-training process in order to investigate how the re-training process affect theperformance of no-reference image quality metric for face biometric images. The experimental resultsshow that the performance of the appropriate IQM can be improved for multiple databases, anddifferent re-training setups can influence the IQM's performance

International audience; The fingerprint is one of the most used biometric modalities because of its persistence, uniqueness characteristics and ease of acquisition. Nowadays, there are large country-sized fingerprint databases for identification purposes, for border access controls and also for Visa issuance procedures around the world. The objective usually is to identify an input fingerprint among a large fingerprint database. In order to achieve this goal, different fingerprint pre-selection, classification or indexing techniques have been developed to speed up the research process to avoid comparison of the input fingerprint template against each fingerprint in the database. Although these methods are fairly accurate for identification process, we think that all of them assume the hypothesis to have a good quality of the fingerprint template for the first step of enrollment. In this paper, we show how the quality of reference templates can impact the performance of identification algorithms. We collect information and implement differents methods from the state of the art of fingerprint identification. Then, for these differents methods, we vary the quality of reference templates by using NFIQ2 metric quality. This allowed us to build a benchmark in order to evaluate the impact of these different enrollment scenarios on identification.

International audience; Biometric authentication systems are increasingly considered in different access control applications. Regarding that users have completely different interactions with these authentication systems, several techniques have been developed in the literature to model distinctive users categories. Doddington zoo is a biometric menagerie that defines and labels user groups with animal species to reflect their behavior with the biometric systems. This menagerie was developed for different biometric modalities including keystroke dynamics. The present study proposes a user dependent adaptive strategy based on the Doddinghton zoo, for the recognition of the user's keystroke dynamics. The novelty of the proposed approach lies in applying an adaptive strategy specific to the characteristics of each user of the Doddinghton zoo menagerie aiming to solve the intra-class variation problems. The obtained results demonstrate competitive performances on significant keystroke dynamics datasets WEBGREYC and CMU.

International audience; With the widespread use of image processing technologies, objective image quality metrics are a fundamental and challenging problem. In this paper, we present a new No-Reference Image Quality Assessment (NR-IQA) algorithm based on visual attention modeling and a multivariate Gaussian distribution to predict the final quality score from the extracted features. Computational modeling of visual attention is performed to compute saliency maps at three resolution levels. At each level, distortions of the input image are extracted and weighted by the saliency maps in order to highlight degradations of visually attracting regions. The generated features are used by a probabilistic model to predict the final quality score. Experimental results demonstrate the effectiveness of the metric and show better performance when compared to well known NR-IQA algorithms.

Cancelable biometric schemes generate secure biometric templates by combining user specific tokens and biometric data. The main objective is to create irreversible, unlinkable, and revocable templates, with high accuracy in matching. In this paper, we cryptanalyze two recent cancelable biometric schemes based on a particular locality sensitive hashing function, index-of-max (IoM): Gaussian Random Projection-IoM (GRP-IoM) and Uniformly Random Permutation-IoM (URP-IoM). As originally proposed, these schemes were claimed to be resistant against reversibility, authentication, and linkability attacks under the stolen token scenario. We propose several attacks against GRP-IoM and URP-IoM, and argue that both schemes are severely vulnerable against authentication and linkability attacks. We also propose better, but not yet practical, reversibility attacks against GRP-IoM. The correctness and practical impact of our attacks are verified over the same dataset provided by the authors of these two schemes., Comment: Some revisions and addition of acknowledgements

Template protection is a crucial issue in biometrics. Many algorithms have been proposed in the literature among secure computing approaches, crypto-biometric algorithm and feature transformation schemes. The BioHashing algorithm belongs to this last category and has very interesting properties. Among them, we can cite its genericity since it could be applied on any biometric modality, the possible cancelability of the generated BioCode and its efficiency when the secret is not stolen by an impostor. Its main drawback is its weakness face to a combined attack (false acceptance with the stolen secret scenario). In this paper, we propose a transformation-based biometric template protection scheme as an improvement of the BioHashing algorithm where the projection matrix is generated by combining the secret and the biometric data. Experimental results on three biometric modalities, namely digital fingerprint, finger knuckle print and hands vein images, show the benefits of the proposed method face to attacks while keeping a good efficiency.

International audience

Many computer graphics applications such as adaptive compression, automatic viewpoint selection, segmentation, etc. use visual saliency to guide their processing. However, all these applications are based on a partial estimation of visual saliency since only the geometric properties of the target mesh are considered leaving aside the colorimetric ones. As human beings, our visual attention is sensitive to both geometric and colorimetric informations. Indeed, colorimetric information influences the movements of the eye when we visualize multimedia content. In this paper, we propose an innovative approach for the detection of global visual saliency that takes into account the geometric and colorimetric properties of the considered mesh, thus simulating the Human Visual System (SVH). To do this, the approach generates two multi-scale saliency maps based on geometric and colorimetric descriptors in the form of patches. These saliency maps are merged using the theory of Evidence. We present the results and benefits of the proposed approach in a process of optimal viewpoint selection., Plusieurs applications de l'informatique graphique telles que la compression adaptative, la sélection automatique de points de vues optimaux, la segmentation, etc. utilisent la saillance visuelle pour guider leurs traitements. Cependant, toutes ces applications reposent sur une estimation partielle de la saillance dans la mesure où seules les propriétés géométriques du maillage cible sont considérées laissant de coté les propriétés colorimétriques. En tant qu'êtres humains, notre attention visuelle est sensible aux informations géométriques et colorimétriques à la fois. En effet, l'information colorimétrique influence les mouvements de l'oeil lorsque nous visualisons un contenu multimédia. Nous proposons dans ce papier une approche novatrice pour la détection de la saillance visuelle globale qui tient compte des propriétés géométriques et colorimétriques du maillage considéré simulant ainsi le Système Visuel Humain (SVH). Pour cela, l'approche génère deux cartes de saillance multi-échelles basées sur des descripteurs géométriques et colorimétriques sous forme de patchs. Ces cartes de saillance sont fusionnées en utilisant la théorie de l'Évidence. Nous présentons les résultats et bénéfices de l'approche proposée dans un processus de séléction automatique de points de vues optimaux.

International audience; With the widespread of computing and mobile devices, authentication using biometrics has received greater attention. Although biometric systems usually provide good solutions, the recognition performance tends to be a ected over time due to changing conditions and aging of biometric data, that results in intra-class variability. Adaptive biometric systems, which adapt the biometric reference over time, have been proposed to deal with such intra-class variability. This paper provides the most up-to-date and complete discussion on adaptive biometrics systems we are aware of, including formalization, terminology, sources or variations that motivates the use of adaptation, adaptation strategies, evaluation methodology and open challenges. This eld of research is sometimes referred to as template update.

Given the shortcomings of traditional networks, software-defined networking (SDN) is considered as the best solution to deal with the constant growth of mobile data traffic. SDN separates the data plane from the control plane, enabling network scalability, and programmability. Initial SDN deployments promoted a centralized architecture with a single controller managing the entire network. This design has proven to be unsuited for nowadays large-scale networks. Though multi-controller architectures are becoming more popular, they bring new concerns. One critical challenge is how to efficiently perform path computation in large networks considering the substantial computational resources needed. This paper proposes HiDCoP, a distributed high-performance control plane for path computation in large-scale SDNs along with its related solutions. HiDCoP employs a hierarchical structure to distribute the load of path computation among different controllers, reducing therefore the transmission overhead. In addition, it uses node parallelism to accelerate the performance of path computation without generating high control overhead. Simulation results show that HiDCoP outperforms existing schemes in terms of path computation time, end-to-end delay, and transmission overhead.

International audience; Smartphones have become ubiquitous in everyday life, storing and generating a huge amount of sensitive personal data which make them vulnerable to increasing security and privacy threats. While protecting smartphones has become a necessity, existing traditional authentication methods, which are mainly PINs and passwords, are facing remarkable drawbacks and behavioural biometrics-based authentication was adopted as the best alternative to ensure better protection. This paper presents a comparative study of many behavioural authentica-tion solutions using smartphone personal communication data. Different approaches are compared such as using Distance Minimization, K-means and Support Vector Machine (SVM) as classification method. The data privacy protection by using the BioHashing algorithm is also considered in the paper. The authentication approaches were tested on a dataset of 93 users with more than 16.000 samples and show promising results with an EER of 10% without any data protection with the One Class SVM method and an EER remarkably lower than 1% for the 3 adopted methods with data privacy protection.

Electric vehicles (EVs) are likely to become very popular within the next few years. With possibly millions of such vehicles operating across the smart cities, smart grid energy providers can be directly impacted by the charging of EV batteries. In order to reduce this impact and optimize energy saving, in this article, we propose a coordinated model for scheduling the plug-in of EVs for charging and discharging energy. The model is based on a new decentralized Fog architecture for smart grid in order to reduce the completion and communication delay of EV energy demand scheduling. To enhance the scheduling of EV demands and predict the future energy flows, we propose a plug-in system of EVs based on calendar planning. We develop a mathematical formalism based on Markov chains using a multipriority queuing theory with cut-off discipline in order to reduce the waiting time to plug-in. We implement three planning algorithms in order to assign priority levels and then optimize the plug-in time into each EV public supply station. To the best of our knowledge, this is the first article that proposes a model that tries to save energy by planning the plug-in of EVs using a cut-off priority queuing model and a decentralized Fog architecture. We evaluate the performances of our solution via extensive simulations using a realistic energy loads from the city of Toronto, and we compare it with other recent works.

Several novel and promising wireless vehicular-network applications have been developed recently. Examples include trip planning, media sharing, and Internet access. However, sufficient network resources [e.g., equipped vehicles, base stations (BSs), roadside units (RSUs), and other infrastructure] to support these applications are not yet available or are deployed at a slow pace. Moreover, these applications require considerable efforts in information gathering and data processing to allow quick decision making and fast feedback to all users. This imposes significant challenges on the development of an efficient wireless vehicularcommunication platform. In this article, we propose a new hierarchical software-defined-network (SDN)-based wireless vehicular fog architecture called a hierarchical SDN for vehicular fog (HSVF) architecture. The HSVF is based on a hybrid SDN control plane that reinforces centralized and distributed management. The proposed SDN control plane includes a trajectory prediction module to mitigate the drawbacks of the frequent handover problem between the RSUs and vehicles. The proposed architecture is then evaluated using a relevant case study that addresses the scheduling of electric vehicle (EV) energy demands.

Development and spread of connected devices, in particular smartphones, requires the implementation of authentication methods. In an ergonomic concern, manufacturers integrates biometric systems in order to deal with logical control access issues. These biometric systems grant access to critical data and application (payment, e-banking, privcy concerns : emails...). Thus, evaluation processes allows to estimate the systems' suitabilty with these uses. In order to improve recognition performances, manufacturer are susceptible to perform multimodal fusion.In this thesis, the evaluation of operationnal biometric systems has been studied, and an implementation is presented. A second contribution studies the quality estimation of speech samples, in order to predict recognition performances.; Le développement et la multiplication de dispositifs connectés, en particulier avec les \textit{smartphones}, nécessitent la mise en place de moyens d'authentification. Dans un soucis d'ergonomie, les industriels intègrent massivement des systèmes biométrique afin de garantir l'identité du porteur, et ce afin d'autoriser l'accès à certaines applications et fonctionnalités sensibles (paiements, e-banking, accès à des données personnelles : correspondance électronique..). Dans un soucis de garantir, une adéquation entre ces systèmes d'authentification et leur usages, la mise en œuvre d'un processus d'évaluation est nécessaire.L'amélioration des performances biométriques est un enjeux important afin de permettre l'intégration de telles solutions d'authentification dans certains environnement ayant d'importantes exigences sur les performances, particulièrement sécuritaires. Afin d'améliorer les performances et la fiabilité des authentifications, différentes sources biométriques sont susceptibles d'être utilisées dans un processus de fusion. La biométrie multimodale réalise, en particulier, la fusion des informations extraites de différentes modalités biométriques.

As a trending method for the authentication, biometrics tends to be integrated in various devices, and in particular in smartphones. If the evaluation is performed on operational device, the biometric sample and algorithm are not reachable by the assessors. So, these latter have to perform an evaluation on a system considered as a black box. This kind of evaluation implies numerous manual comparison. This paper proposes a methodology to perform an evaluation of biometric black boxes. Two preliminary experiments were performed in order to determine an optimized conduct. This paper describes the used methodology to perform evaluation on black boxes systems, and the results obtained on the systems under test.

International audience; Template protection is a crucial issue in biometrics. Many algorithms have been proposed in the literature among secure computing approaches, crypto-biometric algorithm and feature transformation schemes. The BioHashing algorithm belongs to this last category and has very interesting properties. Among them, we can cite its genericity since it could be applied on any biometric modality, the possible cancelability of the generated BioCode and its efficiency when the secret is not stolen by an impostor. Its main drawback is its weakness face to a combined attack (zero effort with the stolen secret scenario). In this paper, we propose a transformation-based biometric template protection scheme as an improvement of the BioHashing algorithm where the projection matrix is generated by combining the secret and the biometric data. Experimental results on two biometric modalities, namely digital fingerprint and finger knuckle print images, show the benefits of the proposed method face to attacks while keeping a good efficiency.

Biometric systems are currently widely used in many applications to control and verify individual's identity. Keystroke dynamics modality has been shown as a promising solution that would be used in many applications such as e-payment and banking applications. However, such systems suffer from several performance limitations (such as cross-devices problem) that prevent their widespread of use in real applications. The objective of this paper is to provide researchers and developers with a public touchscreen-based benchmark collected using a mobile phone and a tablet (both portrait and landscape orientation each). Such a benchmark can be used to assess keystroke-based matching algorithms. Furthermore, It is mainly developed to measure the robustness of keystroke matching algorithms vis-'a-vis cross-devices and orientation variations. An online visualizer for the database is also given to researchers allowing them to visualize the acquired keystroke signals.

National audience

International audience; Biometrics is an emerging technology more and more present in our daily life. However, building biometric systems requires a large amount of data that may be difficult to collect. Collecting such sensitive data is also very time consuming and constrained, s.a. GDPR legislation. In the case of keystroke dynamics, existing databases have less than 200 users. For these reasons, we aim at generating a keystroke dynamics synthetic dataset. This paper presents the generation of keystroke data from known users as a first step towards the generation of synthetic datasets, and could also be used to impersonate users' identity. Results were also found for user generation, but are not part of this paper.

International audience; Regarding the fact that individuals have different interactions with biometric authentication systems, several techniques have been developed in the literature to model different users categories. Doddington Zoo is a concept of categorizing users behaviors into animal groups to reflect their characteristics with respect to biometric systems. This concept was illustrated for different biometric modalities including keystroke dynamics. The present study extends this biometric classification, by proposing a novel adaptive strategy based on the Doddinghton Zoo for user recognition with keystroke dynamics. The obtained results demonstrate competitive performances on significant keystroke dynamics datasets.

The accuracy of face recognition systems is significantly affected by the quality of face sample images. Using degraded face sample images can significantly decrease the performance of face recognition systems. There are not many face quality enhancement methods in the research field. Therefore, it is interesting to investigate whether existing image quality enhancement methods can improve the performance of biometric systems for degraded face images. In this paper, we conduct an experiment to apply nine image enhancement methods to a face database, which contains high quality and degraded face sample images. The experimental results show that some of selected enhancement methods can improve face recognition system performance.

Image quality assessment plays an important role in iris recognition systems because the system performance is affected by low quality iris images. With the development of electronic color imaging, there are more and more researches about visible wavelength (VW) iris recognition. Compared to the near infrared iris images, using VW iris images acquired under unconstrained imaging conditions is a more challenging task for the iris recognition system. However, the number of quality assessment methods for VW iris images is limited. Therefore, it is interested to investigate whether existing no-reference image quality metrics (IQMs) which are designed for natural images can assess the quality of VW iris images. In this paper, we evaluate the performance of 15 selected no-reference IQMs on VW iris biometrics. The experimental results show that several IQMs can assess iris sample quality according to the system performance. © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

In this paper, a novel high speed and efficient area optimal Ate pairing processor implementation over Barreto-Naehrig (BN) and Barreto-Lynn-Scott (BLS12) curves on field-programmable gate array (FPGA) is proposed. The optimal Ate pairing proposed design, based on two steps: Miller loop and final exponentiation, is specifically optimized for FPGA platforms. The Miller Loop and Final Exponentiation algorithms are optimized and modified for careful scheduling to avoid data dependency and to decrease the number of loops and number of temporary variables required for final exponentiation. Furthermore, suitable multiplier combining Toom-Cook and Karatsuba algorithms is proposed to execute the arithmetical computations needed in pairing architecture processor over F p .Therefore, an enhancement in terms of pairing computation speed-up and memory resources capacity management is achieved. In this paper, we select the new pairing parameters, especially that has to be used to ensure the 128-bit security level [1]. The proposed optimal Ate pairing architecture at 128 bits security level has been implemented on Xilinx FPGA Virtex6. Our processor implemented over BN curves achieves the highest reported speed together with the best reported area-time performance on Virtex6 (0.35 ms at 225 Mhz). Finally, for BN-446 and BN-638, the pairing proposed processor performances are estimated.

International audience; We consider the problem of publicly authenticating low entropy data, or even data which vary slightly over time, with an assurance on their freshness at the time of authentication. While traditional digital signatures can be used, it is always interesting to decrease the computation costs of a signature generation. This is particularly true when authentication of a value is requested on a frequent or regular basis. In this paper, we investigate how we can amortize the cost of an asymmetric signature scheme by the use of cryptographic hash-based data structures. For this purpose, we revisit data structures that were previously used for certificate revocation and micro-payment, namely Merkle trees and QuasiModo trees. We thus propose two lightweight signature schemes amortizing the use of a traditional signature scheme. The first one refreshes at low cost the validity period of a signature, assuming we are able to estimate the number of consecutive validity periods over which the data will not change. The second scheme does not require any prediction and refreshes both the validity and the value to authenticate. Finally, optimization trade-offs and the use of a weak hash function are discussed.

The purpose of incremental cryptography is to enable updates of cryptographic forms of documents undergoing modifications, more efficiently than if we had to recompute them from scratch. This paper defines a framework for securely using a variant of the incremental hash function designed by Goi, Siddiqi and Chuah. The condition of use of their hash function is somehow impractical since they assume the blocks of the message are all distinct. In this paper we show how we can discard this strong assumption so as to construct the first practical incremental asymmetric signature scheme that keeps efficient update operations. Finally, as the proposed scheme has the defect to severely expand the signature size, we propose a solution which drastically reduces this drawback.

In this report we present how we used machine learning techniques to detect malicious behaviours in PDF files.At this aim, we first set up a SVM (Support Machine Vector) classifier that was able to detect 99.7% of malware. However, this classifier was easy to lure with malicious PDF, we forged to make them look like clean ones. We first proposed a very naive attack, that was easily stopped by the establishment of a threshold. We also implemented a gradientdescent attack to evade this SVM. This attack was almost 100% successful. In order to fix this problem, we provided counter-measures to the latter attack. A more elaborated features selection, and the use of a threshold, allowed us to stop up to 99.99% of these attacks.Finally, using adversarial learning techniques, we were able to prevent gradient descent attacks by iteratively feeding the SVM with malicious forged PDF. We found that after 3 iterations, every gradient-descent forged PDF were detected, completely preventing the attack.

International audience; Cellular network is facing a severe traffic overload problem caused by the phenomenal growth of mobile data. Offloading part of the mobile data traffic from the cellular network to alternative networks is a promising solution. In this paper, we study the mobile data offloading problem under the architecture of mobile cloud computing, where mobile data can be delivered by WiFi network and device-to-device communication. In order to minimize the overall cost for the data delivery task, it is crucial to reduce cellular network usage while satisfying delay requirements. In our proposed model, we formulate the data offloading task as a finite horizon Markov decision process. We first propose a hybrid offloading algorithm for mobile data with different delay requirements. Moreover, we establish sufficient conditions for the existence of threshold policy. Then, we propose a monotone offloading algorithm based on threshold policy in order to reduce the computational complexity. The simulation results show that the proposed offloading approach can achieve minimal communication cost compared with the other three offloading schemes.

International audience; Securing personal, professional and even official data is a very critical issue nowadays, giving that these infor-mations are safeguarded in different devices (mobile, computer) and various accounts (social networks, e-mails). To protect them from unauthorized acess, users generally are asked to use passwords. But using only these authentication solutions is no longer efficient against hacker attacks. Keystroke dynamics is a biometric promising modality that guarantees the recognition of the user's characteristics; his typing manner on the keyboard. Regarding that the typing rythm of the user changes over time, adaptive biometric solutions help to take into consideration these variations. In this paper we classify user into multiple categories according to Doddonghton Zoo classification. Afterwards, we apply an adaptive strategy specific to each category of users. The achived experiments demonstrate that an update strategy specific to the user class has improved significantly the obtained performances.

The aim of this research is to investigate multi-modality biometric image quality assessment methods for unconstrained samples. Studies of biometrics noted the significance of sample quality for a recognition system or a comparison algorithm because the performance of the biometric system depends mainly on the quality of the sample images. The need to assess the quality of multi-modality biometric samples is increased with the requirement of a high accuracy multi-modality biometric systems. In this thesis we describe the work carried out to use no-reference image quality metrics for the assessment of multi-modality biometric sample images. The intended behavior of such metrics is to evaluate or predict the quality of natural image as human observers would perceive it. Following an introduction and background in biometrics and biometric sample quality, we introduce the concept of biometric sample quality assessment for multiple modalities. Recently established ISO/IEC quality standards for fingerprint, iris, and face are presented. In addition, sample quality assessment approaches which are designed specific for contact-based and contactless fingerprint, nearinfrared-based iris and visible wavelength iris, as well as face are surveyed. Following the survey, approaches for the performance evaluation of biometric sample quality assessment methods are also investigated. Based on the knowledge gathered from the biometric sample quality assessment challenges, we propose a common framework for the assessment of multi-modality biometric image quality. We review the previous classification of image-based quality attributes for a single biometric modality and investigate what are the common image-based attributes for multi-modality. Then we select and re-define the most important image-based quality attributes for the common framework. In order to link these quality attributes to the real biometric samples, we develop a new multi-modality biometric image quality database which has both high quality sample images and degraded images for contactless fingerprint, visible wavelength iris, and face modalities. The degradation types are based on the selected common image-based quality attributes. Another important aspect in the proposed common framework is the image quality metrics and their applications in biometrics. We first introduce and classify the existing image quality metrics and then conducted a brief survey of no-reference image quality metrics, which can be applied to biometric sample quality assessment. Plus, we investigate how no-reference image quality metrics have been used for the quality assessment for fingerprint, iris, and face biometric modalities. From the fingerprint quality ISO/IEC standard we already know what kind of degradations that can affect the performance of a fingerprint recognition system, however, it is not clear if the system performance will be influenced, or to what extent. Therefore, an experiment is conducted to investigate how a number of selected image degradations influence the performance of a fingerprint recognition system, and, influence the performance of a fingerprint quality assessment approach. According to the availability of open source biometric recognition applications, the existing contact-based fingerprint recognition algorithms have difficulties to extract the minutiae from the high quality contactless fingerprint images from our database. Therefore, we review 10 existing methods for the enhancement of contact-based fingerprint and select two suitable methods to verify whether they can be applied to contactless fingerprint. Finally, we propose an improved 3-step contactless fingerprint image enhancement approach for the minutia extraction. However, after the image enhancement process, the number of detected minutiae is still under an acceptable level. We decide to leave contactless fingerprint for our further study. The experiments for the performance evaluation of no-reference image quality metrics for visible wavelength face and iris modalities are conducted. The experimental results indicate that there are several no-reference image quality metrics that can assess the quality of both iris and face biometric samples. Lastly, we optimize the best metric by re-training it. The re-trained image quality metric can provide better recognition performance than the original. Through the work carried out in this thesis we have shown the applicability of no-reference image quality metrics for the assessment of unconstrained multi-modality biometric samples. Sammendrag - Målet med denne forskningen er å undersøke biometrisk bildekvalitet med flere modaliteter i ukontrollerte omgivelser. Tidligere forsknings innen biometri har påpekt betydningen av bildekvalitet for et gjenkjennelsessystem eller sammenligningsalgoritmer fordi ytelsen til det biometriske systemet avhenger hovedsakelig av kvaliteten av bildene. Behovet for å vurdere kvaliteten på multimodale biometriske bilder øker med kravet om høy nøyaktighet i multimodale biometriske systemer. I denne avhandlingen beskriver vi arbeidet som utføres for å gjøre beregninger av bildekvalitet uten referansebilder for vurdering av biometriske bilder med flere modaliteter. Den tiltenkte oppgaven til slike beregninger er å evaluere eller forutsi kvaliteten på det naturlige bildet som menneskelige observatører ville oppleve det. Etter en introduksjon og bakgrunn i biometri og biometrisk kvalitet, presenterer vi begrepet biometrisk bildekvalitet vurdering for flere modaliteter. Nylig etablerte ISO / IEC kvalitetsstandarder for fingeravtrykk, iris og ansikt blir presentert. I tillegg tilnærmes bildekvalitetsvurdering som er konstruert spesifikt for kontaktbaserte og kontaktløse fingeravtrykk, nærinfrarødbasert iris og synlig bølgelengde iris, samt ansikt blir undersøkt. Etter dette, undersøker vi ytelsesevaluering av biometriske bildekvalitetsmetoder. Basert på kunnskap samlet fra dagens utfordringer innen biometriske bildekvalitetsevaluering, foreslår vi et felles rammeverk for vurdering av multimodal biometrisk bildekvalitet. Vi går igjennom tidligere klassifisering av bildebaserte kvalitetsattributter for hver biometrisk modalitet og undersøker hva som er vanlig bildebaserte attributter for multimodale biometriske systemer. Basert på dette velger vi og definerer de viktigste bildebaserte kvalitetsattributter for det felles rammeverket. For å knytte disse kvalitetsattributene til de virkelige biometriske bildene utvikler vi en ny biometrisk bildekvalitetsdatabase med flere modaliteter, som har både høy kvalitets bilder og forringede bilder for kontaktløse fingeravtrykk, synlig bølgelengde iris og ansikt. Forringede bilder er basert på vanlige bildebaserte kvalitetsattributter. Et annet viktig aspekt i det foreslåtte rammeverket er bildekvalitetsmålinger og deres applikasjoner i biometri. Først introduserer og klassifiserer vi eksisterende bildekvalitetsmetrikker og deretter viser vi en oversikt over ikke-referanse bildekvalitetsmetrikker uten referansebilde som kan brukes til biometrisk bildekvalitetsvurdering. I tillegg undersøker vi hvordan bildekvalitetsmetrikker har blitt brukt til kvalitetsvurdering for fingeravtrykk, iris og ansikt. Fra ISO / IEC-standarden for fingeravtrykkskvalitet vet vi allerede hva slags forringelse av bildene som kan påvirke ytelsen til et fingeravtrykksgjenkjenningssystem. Det er imidlertid ikke klart om systemytelsen vil bli påvirket, eller i hvilken utstrekning. Derfor utføres et eksperiment for å undersøke hvordan et antall valgte bildeforringelser påvirker ytelsen til et fingeravtrykksgjenkjenningssystem, og sammenhengen mellom kvalitetsvurdering av bildene og ytelsen av systemet. I henhold til tilgjengeligheten av åpen kildekode biometrisk anerkjennelsesprogrammer, har de eksisterende kontaktbaserte fingeravtrykksgjenkjenningsalgoritmene problemer med å trekke ut referansepunkter fingeravtrykksbildene med høy kvalitet fra vår database. På grunn av dette tester vi 10 eksisterende metoder for forbedring av kontaktbasert fingeravtrykk, og velger to egnede metoder for å kontrollere om de kan brukes til kontaktløse fingeravtrykk. Til slutt foreslår vi et forbedret 3-trinns bildeforbedringsmetode for kontaktløse fingeravtrykk. Imidlertid, etter bildeforbedringsprosessen, er antall referansepunkter fortsatt under et akseptabelt nivå. Vi bestemmer oss for å ikke fortsette forskningen på kontaktløse fingeravtrykk i våre videre studier. Resten av avhandlingen omfatter eksperimenter for ytelsesevaluering av kvalitetsmetrikker uten referansebilde for synlig bølgelengde ansiktsbilder og iris. De eksperimentelle resultatene indikerer at det er flere bildekvalitetsmetrikker som kan vurdere kvaliteten på både iris- og ansiktsbilder. Til slutt optimaliserer vi den beste metrikken, og den optimaliserte metrikken gir bedre ytelse enn originalen. Gjennom arbeidet i denne avhandlingen har vi vist anvendeligheten av bildekvalitetsmetrikker for vurdering av biometriske bilder med flere modaliteter. Résumé - L’objectif de ces travaux de recherche est d’étudier les méthodes d’évaluation de la qualité des images biométriques multimodales sur des échantillons acquis de manière non contrainte. De nombreuses s études ont noté l’importance de la qualité de l’échantillon pour un système de reconnaissance ou un algorithme de comparaison, puisque la performance du système biométrique est intrinsèquement dépendant de la qualité des images de l’échantillon. Dès lors, la nécessité d’évaluer la qualité des échantillons biométriques pour plusieurs modalités (empreintes digitales, iris, visage, etc.) est devenue primordiale notamment avec l’apparition de systèmes biométriques multimodaux de haute précision. Dans cette thèse, nous décrivons le travail effectué pour étudier l’adaptabilité des métriques de qualité d’image sans référence pour l’évaluation d’images biométriques de plusieurs modalités. Le comportement attendu de ces métriques est d’évaluer ou de prédire la qualité de l’image naturelle telle que le font les observateurs humains. Après une introduction présentant un historique de la biométrie et des préceptes liés à la qualité des échantillons biométriques, nous présentons le concept d’évaluation de la qualité des échantillons pour plusieurs modalités. Les normes de qualité ISO / CEI récemment établies pour les empreintes digitales, l’iris et le visage sont présentées. De plus, des approches d’évaluation de la qualité des échantillons conçues spécifiquement pour les empreintes digitales avec et sans contact, pour l’iris (dont une image est capturée en proche infrarouge et dans le domaine visible), ainsi que le visage sont étudiées. Finalement, des techniques d’évaluation des performances des mesures de qualité des échantillons biométriques sont également étudiées. Sur la base des conclusions formulées suite à l’étude des solutions algorithmiques portant sur l’évaluation de la qualité des échantillons biométriques, nous proposons un cadre commun pour l’évaluation de la qualité d’image biométrique pour plusieurs modalité. Après avoir étudié les attributs de qualité basés sur l’image par modalité biométrique, nous examinons quelle intersection existe pour l’ensemble des modalités. Ensuite, nous sélectionnons et redéfinissons les attributs de qualité basés sur l’image qui sont les plus importants afin de définir un cadre commun. Afin de relier ces attributs de qualité aux vrais échantillons biométriques, nous développons une nouvelle base de données de qualité d’image biométrique multi-modalité qui contient des images échantillons de haute qualité et des images dégradées pour l’empreinte digitale acquise sans contact, l’iris (dont l’acquisition est réalisée dans le spectre visible) et le visage. Les types de dégradation appliqués sont liés aux attributs de qualité qui sont communs aux diverses modalités et qui sont basés sur l’image. Un autre aspect important du cadre commun proposé est la qualité de l’image et ses applications en biométrie. Nous avons d’abord introduit et classifié les métriques de qualité d’image existantes, puis effectué un bref aperçu des métriques de qualité d’image sans référence, qui peuvent être appliquées pour l’évaluation de la qualité des échantillons biométriques. De plus, nous étudions comment les mesures de qualité d’image sans référence ont été utilisées pour l’évaluation de la qualité des empreintes digitales, de l’iris et des modalités biométriques du visage. D’après la norme ISO / CEI portant sur la qualité des empreintes digitales, nous connaissons les types de dégradations qui peuvent affecter les performances d’un système de reconnaissance d’empreintes digitales. Cependant, il n’est pas clair dans quelle mesure les performances du système sont influencées. Par conséquent, une expérience est menée pour étudier de quelle manière les dégradations appliquées sur les images influencent les performances d’un système de reconnaissance d’empreintes digitales, et dans quelle mesure elles influencent les performances des mesures de qualité des empreintes digitales. Les algorithmes existants de reconnaissance d’empreintes digitales acquises par contact ne sont pas nécessairement performants lors de la phase d’extraction des minuties sur les images d’empreintes digitales acquises sans contact et de haute qualité présentes dans notre base de données. Par conséquent, dix méthodes d’amélioration des images d’empreinte digitale acquises avec contact extraites de l’état de l’art sont étudiées. Seules les deux meilleures méthodes dont sélectionnées afin vérifier dans quelle mesure elles sont applicables pour des images d’empreinte digitale acquises sans contact. Finalement, nous proposons une approche d’amélioration des images d’empreinte digitale acquises sans contact pour l’extraction de minuties. Cependant, après application du processus d’amélioration d’image, le nombre de minuties détectées est toujours inférieur à un niveau acceptable. Ceci fera l’objet de travaux ultérieurs afin d’aboutir à un processus d’extraction de minuties cohérent. Des expériences pour l’évaluation de la performance des métriques de qualité d’image sans référence sur les images de visage et de l’iris sont effectuées. Les résultats expérimentaux indiquent qu’il existe plusieurs métriques qui peuvent évaluer la qualité des échantillons biométriques de l’iris et du visage avec un fort coefficient de correlation. La méthode obtenant les meilleurs résultats en terme de performance est ré-entrainée sur des images d’empreintes digitales, ce qui permet d’augmenter significativement les performances du système de reconnaissance biométrique. À travers le travail réalisé dans cette thèse, nous avons démontré l’applicabilité des métriques de qualité d’image sans référence pour l’évaluation d’échantillons biométriques multi-modalité non contraints. Xinwei Liu carried out his PhD work in cooperation with another university by agreement on joint supervision and doctoral education (cotutelle agreement). Liu carried out his PhD work at the Department of Computer Science, NTNU and at the Université de Caen Normandie, France. Liu’s main supervisors have been Professor Marius Pedersen, NTNU and Professor Christophe Charrier, University of Caen Normandy. Co-supervisor has been Professor Patrick Bours, NTNU.

International audience; Internet is increasingly used for card not present e-commerce ar-chitectures. Several protocols, such as 3D-Secure, have been proposed in the literature by Card schemes or academics. Even if some of them are deployed in real life, these solutions are not perfect considering data security and user's privacy. In this paper, we present a comparative study of existing solutions for card not present e-commerce solutions. We consider the main security and privacy trends of e-payment in order to make an objective comparison of existing solutions. This comparative study illustrates the need to consider privacy in deployed e-commerce architectures. This has never been more urgent with the recent release of the new specifications of 3D-secure.

National audience

International audience; Password based applications are commonly used in our daily lives such as the social networks, e-mails, e-commerce, and e-banking. Given the increasing number of hacker attacks, the only use of passwords is not enough to protect personal data and does not meet usability requirements. Keystroke dynamics is a promising solution that decreases the vulnerability of passwords to guessing attacks by analyzing the typing manner of the user. Despite its efficiency in the discrimination between users, it remains non-industrialized essentially due to the tedious learning phase and the intra-class variation of the users' characteristics. In this paper, we propose a double serial mechanism to adapt the user's model over time. An important property of the proposed solution relies in its usability as we only use a single sample as user's reference during the account creation. We demonstrate that the proposed method offers competitive performances while keeping a high usability.

International audience

International audience; Fake identities and identity theft are issues whose relevance is increasing in the social network domain. This paper deals with this problem by proposing an innovative approach which combines a collaborative mechanism implementing a trust graph with keystroke-dynamic-recognition techniques to trust identities. The trust of each node is computed on the basis of neighborhood recognition and behavioral biometric support. The model leverages the word of mouth propagation and a settable degree of redundancy to obtain robustness. Experimental results show the benefit of the proposed solution even if attack nodes are present in the social network.

The quality assessment of biometric data proceeds as a toll to decide whether a biometric sample may be used to generate the user's reference template. Many studies showed its signi?cant impact on the subsequent performance of the biometric system. Since many metrics are proposed for this purpose by researchers or standardization institutions, their relevance should be studied in particular to evaluate their relative usefulness. This paper provides a comparative study of fingerprint quality assessment (FQA) metrics. We consider the enrollment selection validation approach to perform an objective comparison of them. We show the efficiency of 7 well known FQA metrics on 9 datasets. Results show a dependency of these metrics to the dataset (i.e. the fingerprint sensor) and for matchers a similar behavior.

Our visual attention is attracted by specific areas into 3D objects (represented by meshes). This visual attention depends on the degree of saliency exposed by these areas. In this paper, we propose a novel multi-scale approach for detecting salient regions. To do so, we define a local surface descriptor based on patches of adaptive size and filled in with a local height field. The single-scale saliency of a vertex is defined as its degree measure in the mesh with edges weights computed from adaptive patch similarities weighted by the local curvature. Finally, the multi-scale saliency is defined as the average of single-scale saliencies weighted by their respective entropies. The contribution of the multi-scale aspect is analyzed and showed through the different results. The strength and the stability of our approach with respect to noise and simplification are also studied. Our approach is compared to the state-of-the-art and presents competitive results. Graphical abstractDisplay Omitted HighlightsWe model the surface of the mesh by a non oriented graph. We associate to each vertex a normal vector and a 2D tangent plane.To describe the mesh surface, we use local patches with adaptive size. These patches are filled with a local height field of the spherical neighborhood.The single-scale saliency is defined as the vertices's degree measures with edges weights accounting for vertices similarities.The multi-scale saliency is defined as the average of single-scale saliencies weighted by their respective entropies.We discuss, compare the results with the state-of-the-art, and demonstrate the stability of our approach. We also present an application for selecting the most important viewpoints of 3D meshes.

International audience

International audience; Various applications on Internet require information on users, to verify their right to access services (verification of identity proofs s.a. passwords), to avoid attacks (s.a. paedophilia, profile usurpations), or to give trust to users (e.g. in social networks). In this paper, we introduce a method to generate (non-cryptographics) identity-based signatures computed from 1) collection of data from user biomet-rics, computer configuration, web browser fingerprinting, 2) data pre-processing, 3) protection of personal information through generation of a binary code (our signature). We illustrate the benefits of the proposed method with preliminary results on real personal information.

National audience; De nombreuses applications sur Internet nécessitent d'avoir des informations sur l'internaute, pour vérifier qu'il a bien le droit d'accéder à un service numérique (vérification d'une preuve d'identité comme un mot de passe), pour éviter des attaques (pédopornographie, usurpation de profil, ...) ou pour donner de la confiance aux autres utilisateurs (réseaux sociaux). Néanmoins, la protection législative des données personnelles (notamment dans le cadre Européen du Règlement général sur la protection des données) ne permet pas une collecte portant atteinte à la vie privée de l'internaute. Nous proposons donc dans cet article une méthode de génération d'une signature non-cryptographique basée sur l'identité de l'individu. Cette approche se veut respectueuse de la vie privée par construction et permet d'appréhender plusieurs applications dont l'authentification. Des résultats expérimentaux sont donnés dans ce papier illustrant l'apport de la méthode proposée.

National audience; Le nombre de fonctionnalités offertes par l'IOT croît fortement.Dans ce papier, nous mettons en évidence les problèmes de sécurité engendrés par cette tendance. Pour ce faire, nous proposons une mise en relation des fonctionnalités principales des objets de l'IoT avec une taxonomie des menaces. Finalement, nous introduisons une méthodologie pour limiter les menaces passant par un contrôle des fonctionnalités.

International audience; Unlock patterns are used for authentication in mobile smart devices, yet they are vulnerable to attacks, since only the pattern draw is required. Extra biometric data of the user while drawing the unlock pattern passwords may strengthen the authentication, such as the speed of drawing, the pressure of the finger on the touch screen. Such biometric modality is referred to as behavioral biometrics. Besides, voice is also a behavioral biometric modality, as well as a physiological one. Hence, statistical models such as Gaussian mixture models (GMM) with universal background modeling (UBM) are widely used in speaker verification systems. In this work, we propose to apply and adapt a framework usually dedicated to speaker verification to recognize the unlock patterns based on users' behavior. We evaluate the performance using equal error rate for different combinations of features and varying number of mixtures. As a result of the combination of features, an equal error rate as low as 9.25% on average is obtained, which is promising for a preliminary study on GMM-UBM applied to unlock pattern based biometric recognition.

International audience

International audience; Color provides important information and features for face recognition. Different color spaces possess different characteristics and are suitable for different applications. In this paper, we propose to investigate how different color space components influence the performance of degraded face recognition. Towards this goal, nine color space components are selected for the evaluation. In addition, four different types of image-based degradations are applied to face image samples in order to discover the impact on the performance of face recognition. The experimental results show that, all selected color components have similar influence to the performance of face recognition system depend on the acquisition devices and the experimental setups.

National audience

International audience