This article presents SVC-onGoing1, an on-going competition for on-line signature verification where researchers can easily benchmark their systems against the state of the art in an open common platform using large-scale public databases, such as DeepSignDB2 and SVC2021_EvalDB3, and standard experimental protocols. SVC-onGoing is based on the ICDAR 2021 Competition on On-Line Signature Verification (SVC 2021), which has been extended to allow participants anytime. The goal of SVC-onGoing is to evaluate the limits of on-line signature verification systems on popular scenarios (office/mobile) and writing inputs (stylus/finger) through large-scale public databases. Three different tasks are considered in the competition, simulating realistic scenarios as both random and skilled forgeries are simultaneously considered on each task. The results obtained in SVC-onGoing prove the high potential of deep learning methods in comparison with traditional methods. In particular, the best signature verification system has obtained Equal Error Rate (EER) values of 3.33% (Task 1), 7.41% (Task 2), and 6.04% (Task 3). Future studies in the field should be oriented to improve the performance of signature verification systems on the challenging mobile scenarios of SVC-onGoing in which several mobile devices and the finger are used during the signature acquisition.

Thanks to Mr. James Bond, we are aware that diamonds are forever but, are fingerprints? It is well known that biometrics brings to the security field a new paradigm; unlike traditional systems, individuals are not identified by something that they have or they know, but by what they are. While such an approach entails some clear advantages, an important question remains: is what we are today the same as what we will be tomorrow? This paper addresses such a key problem in the fingerprint modality based on a database of over 400K impressions coming from more than 250K different fingers. The database was acquired under real operational conditions and contains fingerprints from subjects aged 0-25 and 65-98 years. Fingerprint pairs were collected with a time difference that ranges between 0 and 7 years. Such a unique set of data has allowed us to analyze both the age and ageing effects, shedding some new light into issues, such as fingerprint permanence and fingerprint quality.

We study user interaction with touchscreens based on swipe gestures for personal authentication. This approach has been analyzed only recently in the last few years in a series of disconnected and limited works. We summarize those recent efforts and then compare them to three new systems (based on support vector machine and Gaussian mixture model using selected features from the literature) exploiting independent processing of the swipes according to their orientation. For the analysis, four public databases consisting of touch data obtained from gestures sliding one finger on the screen are used. We first analyze the contents of the databases, observing various behavioral patterns, e.g., horizontal swipes are faster than vertical independently of the device orientation. We then explore an intra-session scenario, where users are enrolled and authenticated within the same day, and an inter-session one, where enrollment and test are performed on different days. The resulting benchmarks and processed data are made public, allowing the reproducibility of the key results obtained based on the provided score files and scripts. In addition to the remarkable performance, thanks to the proposed orientation-based conditional processing, the results show various new insights into the distinctiveness of swipe interaction, e.g., some gestures hold more user-discriminant information, data from landscape orientation is more stable, and horizontal gestures are more discriminative in general than vertical ones.

The wide deployment of biometric recognition systems in the last two decades has raised privacy concerns regarding the storage and use of biometric data. As a consequence, the ISO/IEC 24745 international standard on biometric information protection has established two main requirements for protecting biometric templates: irreversibility and unlinkability. Numerous efforts have been directed to the development and analysis of irreversible templates. However, there is still no systematic quantitative manner to analyze the unlinkability of such templates. In this paper, we address this shortcoming by proposing a new general framework for the evaluation of biometric templates' unlinkability. To illustrate the potential of the approach, it is applied to assess the unlinkability of the four state-of-the-art techniques for biometric template protection: biometric salting, bloom filters, homomorphic encryption, and block re-mapping. For the last technique, the proposed framework is compared with other existing metrics to show its advantages.

One of the fields that still today remains largely unexplored in biometrics is 3D fingerprint recognition. This gap is mainly explained by the lack of scanners capable of acquiring on a touchless, fast, reliable and repeatable way, accurate fingerprint 3D spatial models. As such, full 3D fingerprint data with which to produce research and advance this field is almost nonexistent. If such acquisition process was possible, it could represent the beginning of a real paradigm shift in the way fingerprint recognition is performed. The present paper is a first promising step to address the fascinating challenge of 3D fingerprint acquisition and recognition. It presents a new full 3D touchless fingerprint scanner, a new database with 1,000 3D finger-print models, a new segmentation method based on the additional spatial information provided by the models, and initial baseline verification results.

Traditionally, the accuracy of signature verification systems has been evaluated following a protocol that considers two independent impostor scenarios: random forgeries and skilled forgeries. Although such an approach is not necessarily incorrect, it can lead to a misinterpretation of the results of the assessment process. Furthermore, such a full separation between both types of impostors may be unrealistic in many operational real-world applications. The current article discusses the soundness of the random-skilled impostor dichotomy and proposes complementary approaches to report the accuracy of signature verification systems, discussing their advantages and limitations.

After more than two decades of research in the field of password strength estimation, one clear conclusion may be drawn: no password strength metric by itself is better than all other metrics for every possible password. Building upon this certainty and also taking advantage of the knowledge gained in the area of information fusion, in this paper, we propose a novel multimodal strength metric that combines several imperfect individual metrics to benefit from their strong points in order to overcome many of their weaknesses. The final multimodal metric comprises different modules based both on heuristics and statistics, which, after their fusion, succeed to provide in real time a realistic and reliable feedback regarding the “guessability” of passwords. The validation protocol and the test results are presented and discussed in a companion paper.

A novel multimodal method for the estimation of password strength was presented in Part I of this series of two papers. In this paper, the experimental framework used for the evaluation of the novel approach is described. The method is evaluated following a reproducible protocol, which includes a three-dimensional approach: 1) deterministic assessment; 2) statistical assessment; and 3) third parties assessment (thanks to the availability upon request of an executable application that integrates the multimodal meter). The key experiment of the protocol compares, from a probabilistic point of view, the strength distributions assigned to passwords broken with increasingly complex attacking approaches, following a common strategy in a typical password cracking session. The experimental evaluation is carried out not only for the new meter, but also for other strength estimators from the state of the art, comparing their overall performance. In addition to its consistent results, the proposed method is highly flexible and can be adjusted to specific environments or to a certain password policy. Furthermore, it can also evolve over time in order to naturally adjust to new password selection trends followed by users.

Biometric systems typically suffer a significant loss of performance when the acquisition sensor is changed between enrolment and authentication. Such a problem, commonly known as sensor interoperability, poses a serious challenge to the accuracy of matching algorithms. The present work addresses for the first time the sensor interoperability issue in 3D face recognition systems, analysing the performance of two popular and well known techniques for 3D facial authentication. For this purpose, a new gender-balanced database comprising 3D data of 26 subjects has been acquired using two devices belonging to the new generation of low-cost 3D sensors. The results show the high sensor-dependency of the tested systems and the need to develop matching algorithms robust to the variation in the sensor resolution.

User authentication using simple gestures is now common in portable devices. In this work, authentication with free-form sketches is studied. Verification systems using dynamic time warping and Gaussian mixture models are proposed, based on dynamic signature verification approaches. The most discriminant features are studied using the sequential forward floating selection algorithm. The effects of the time lapse between capture sessions and the impact of the training set size are also studied. Development and validation experiments are performed using the DooDB database, which contains passwords from 100 users captured on a smartphone touchscreen. Equal error rates between 3% and 8% are obtained against random forgeries and between 21% and 22% against skilled forgeries. High variability between capture sessions increases the error rates.

On-line signature verification still remains a challenging task within biometrics . Due to their behavioural nature (opposed to anatomic biometric traits), signatures present a notable variability even between successive realizations. This leads to higher error rates than other largely used modalities such as iris or fingerprints and is one of the main reasons for the relatively slow deployment of this technology. As a step towards the improvement of signature recognition accuracy , the present paper explores and evaluates a novel approach that takes advantage of the performance boost that can be reached through the fusion of on-line and off-line signatures. In order to exploit the complementarity of the two modalities, we propose a method for the generation of enhanced synthetic static samples from on-line data. Such synthetic off-line signatures are used on a new on-line signature recognition architecture based on the combination of both types of data: real on-line samples and artificial off-line signatures synthesized from the real data. The new on-line recognition approach is evaluated on a public benchmark containing both real versions (on-line and off-line) of the exactly same signatures. Different findings and conclusions are drawn regarding the discriminative power of on-line and off-line signatures and of their potential combination both in the random and skilled impostors scenarios.

One of the main challenges of off-line signature verification is the absence of large databases. A possible alternative to overcome this problem is the generation of fully synthetic signature databases, not subject to legal or privacy concerns. In this paper we propose several approaches to the synthesis of off-line enhanced signatures from real dynamic information. These synthetic samples show a performance very similar to the one offered by real signatures, even increasing their discriminative power under the skilled forgeries scenario, one of the biggest challenges of handwriting recognition. Furthermore, the feasibility of synthetically increasing the enrolment sets is analysed, showing promising results.

A new face anti-spoofing method based on general image quality assessment is presented. The proposed approach presents a very low degree of complexity which makes it suitable for real-time applications, using 14 image quality features extracted from one image (i.e., the same acquired for face recognition purposes) to distinguish between legitimate and impostor samples. The experimental results, obtained on two publicly available datasets, show very competitive results compared to other state-of-the-art methods tested on the same benchmarks. The findings presented in the work clearly suggest that the analysis of the general image quality of real face samples reveals highly valuable information that may be very efficiently used to discriminate them from fake images.

Biometric data are considered sensitive personal data and any privacy leakage poses severe security risks. Biometric templates should hence be protected, obscuring the biometric signal in a non-reversible manner, while preserving the unprotected system's performance. In the present work, irreversible face templates based on adaptive Bloom filters are proposed. Experiments are carried out on the publicly available Bio Secure DB utilizing the free Bob image processing toolbox, so that research is fully reproducible. The performance and security evaluations proof the irreversibility of the protected templates, while preserving the verification performance. Furthermore, template size is considerably reduced.

One of the biggest challenges faced by law enforcement entities in the present digital era, is fighting against online Child Sexual Abuse (CSA), due in particular to the massive amount of data that they receive for analysis. Pattern recognition system can provide an aid, e.g., to ease the identification of both the perpetrator and the victim of the crime. In particular, ancillary cues related the identity of the involved persons, like age, race or gender, can represent a significant aid for identification. These cues can be estimated using statistical classifiers on face features. In this work, we explore one of these ancillary cues, namely the gender. The research community has provided methods for gender recognition able to achieve good performance with adults. However, in the case of CSA, victims are minors (typically, very young children). Children gender recognition may be difficult even for humans, due to the lack of many gender-specific face traits usually present in adult faces. Totally uncontrolled poses and illumination conditions, that might be found in CSA material, represent an additional issue. We propose to tackle this problem by the use of contextual information to complement face features used by traditional algorithms. In particular, we exploit the image context of the face, that is, the portion of the image surrounding the face. This is motivated by the usage that humans themselves make of face external information, such as the hair or earrings, to take decisions on this task. The proposed approach is tested on a novel data base of faces of children, collected from royalty-free stock-photography web sites, which show totally unconstrained conditions. The reported results are promising and set the way for a deeper study of the use of the face context for estimating ancillary identification cues.

To ensure the actual presence of a real legitimate trait in contrast to a fake self-manufactured synthetic or reconstructed sample is a significant problem in biometric authentication, which requires the development of new and efficient protection measures. In this paper, we present a novel software-based fake detection method that can be used in multiple biometric systems to detect different types of fraudulent access attempts. The objective of the proposed system is to enhance the security of biometric recognition frameworks, by adding liveness assessment in a fast, user-friendly, and non-intrusive manner, through the use of image quality assessment. The proposed approach presents a very low degree of complexity, which makes it suitable for real-time applications, using 25 general image quality features extracted from one image (i.e., the same acquired for authentication purposes) to distinguish between legitimate and impostor samples. The experimental results, obtained on publicly available data sets of fingerprint, iris, and 2D face, show that the proposed method is highly competitive compared with other state-of-the-art approaches and that the analysis of the general image quality of real biometric samples reveals highly valuable information that may be very efficiently used to discriminate them from fake traits.

Proceedings of Interspeech 2013, Lyon (France)

An analysis of the quality of on-line handwritten signatures is carried out based on the Sigma-Lognormal model. In the study, two main issues are addressed from a kinematic perspective of humanly-produced movements. On the one hand, what makes some signatures perform better than others in automatic signature verification systems, and on the other hand if that information may be used as a quality measure in order to predict the expected performance of a given sample. Experiments were carried out on the MCYT database and show the high potential of certain kinematic features for signature quality assessment.

The kinematical information present in synthetically generated signatures is analyzed using the Sigma-Lognormal model and compared to the kinematical properties of real samples. Experiments are carried out on totally independent development and test sets and show a high degree of similarity between humanly produced and artificial signatures. One particular flaw is found in the velocity profile of synthetic signatures. Two possible solutions are proposed to improve the synthetic generation method using the Kinematic Theory of rapid human movements.

Doodle-based graphical passwords represent a challenging scenario due to their high variability and the tendency to be graphically simple. Despite this, doodle-based authentication using touch screens is a promising lightweight user verification method. Several works have been published in this field, although they report in general experimental verification results over small and private databases. In this paper we analyze the performance of several state-of-the-art systems for doodle verification, using the recently acquired DooDB database, which is publicly available. Several algorithms are tested, from the fields of gesture recognition and doodle and signature verification. A comparative study of their performance is done, and future research directions are pointed out.

A new multimodal biometric database designed and acquired within the framework of the European BioSecure Network of Excellence is presented. It is comprised of more than 600 individuals acquired simultaneously in three scenarios: 1) over the Internet, 2) in an office environment with desktop PC, and 3) in indoor/outdoor environments with mobile portable hardware. The three scenarios include a common part of audio/video data. Also, signature and fingerprint data have been acquired both with desktop PC and mobile portable hardware. Additionally, hand and iris data were acquired in the second scenario using desktop PC. Acquisition has been conducted by 11 European institutions. Additional features of the BioSecure Multimodal Database (BMDB) are: two acquisition sessions, several sensors in certain modalities, balanced gender and age distributions, multimodal realistic scenarios with simple and quick tasks per modality, cross-European diversity, availability of demographic data, and compatibility with other multimodal databases. The novel acquisition conditions of the BMDB allow us to perform new challenging research and evaluation of either monomodal or multimodal biometric systems, as in the recent BioSecure Multimodal Evaluation campaign. A description of this campaign including baseline results of individual modalities from the new database is also given. The database is expected to be available for research purposes through the BioSecure Association during 2008.

In this paper, we study the impact of an incremental level of skill in the forgeries against signature verification systems. Experiments are carried out using both off-line systems, involving the discrimination of signatures written on a piece of paper, and on-line systems, in which dynamic information of the signing process (such as velocity and acceleration) is also available. We use for our experiments the BiosecurID database, which contains both on-line and off-line versions of signatures, acquired in four sessions across a 4 month time span with incremental level of skill in the forgeries for different sessions. We compare several scenarios with different size and variability of the enrolment set, showing that the problem of skilled forgeries can be alleviated as we consider more signatures for enrolment.

A brute force attack using synthetically generated handwritten signatures is performed against a HMM-based signature recognition system. The generation algorithm of synthetic signatures is based on the spectral analysis of the trajectory functions and has proven to produce very realistic results. The experiments are carried out by attacking real signature models from the MCYT database (which comprises 8,250 signature samples from 330 users). Results show that such an attack is feasible, thus arising the necessity of introducing countermeasures against this type of vulnerability in real applications.

A novel scheme to generate multiple synthetic samples from a real on-line handwritten signature is proposed. The algorithm models a transmission channel which introduces a certain distortion into the real signature to produce the different synthetic samples. The method is used to increase the amount of data of the clients enrolling on a state-of-the-art HMM-based signature verification system. The enhanced enrollment results in performance improve up to70% between the case in which only one real sample of the user was available for the training, and the case where the proposed algorithm was used to generate additional synthetic training data.

A performance and robustness study for on-line signature verification is presented. Experiments are carried out on the MCYT database comprising 16,500 signatures from 330 subjects, which are parameterized by means of a 100-feature set which can be divided into four different groups according to the signature information they contain, namely: (i) time, (ii) speed and acceleration, (iii) direction, and (iv) geometry. The SFFS feature selection algorithm is used to search for the best performing feature subsets under the skilled and random forgeries scenarios, and to find the most robust subsets against a hill-climbing attack. Comparative experiments are given, where it is shown that the most discriminant parameters are those regarding geometry information, while the most robust are the time related features.

This work reports a preliminary study on the vulnerability evaluation of fingerprint verification systems to direct attacks carried out with fake fingertips created from minutiae templates. The attack is performed by presenting to the acquisition sensor a fake fingertip generated from an image reconstructed from a compromised minutiae-based template. Experiments carried out against a state-of-the-art fingerprint recognition algorithm show that the proposed attack scheme is definitely feasible and highlight a potential security threat in the use of non-encrypted minutiae-based templates.

The proliferation of handheld devices such as PDAs and smart phones represents a new scenario for automatic signature verification. Traditionally, research on signature verification has been carried out employing signatures acquired using digitizing tablets or Tablet-PCs. In this paper we study the effects of the mobile acquisition conditions and we analyze the considerations that must be taken in the new handheld scenario. A signature verification system adapted to handheld devices via feature selection is proposed and a systematic comparison with a traditional pen tablet-based system is performed. The system is combined with another based on hidden Markov models using score fusion. Results confirm an increased signature variability in the case of handheld devices.