With the growing integration of biometric recognition systems into high-security and large-scale deployment scenarios, it is becoming increasingly important to ensure their robustness under realistic operational constraints. This implies designing solutions able to withstand potential adversarial threats that could affect their integrity and accountability, and also taking into account requirements of real-world operating systems such as limited availability of bandwidth and memory for data transmission and storage. Hence, the proposed study deals with presentation attack detection (PAD) for iris recognition, evaluating the effectiveness of Transformer-based frameworks at detecting spoofing attacks relying on fabricated or artificial biometric evidences. More specifically, we focus on the effects of image compression on the quality of iris images and on the resulting PAD performance, considering both traditional techniques such as JPEG as well as next-generation learning-based image codecs such as JPEG AI. We then examine the feasibility of mitigating compression-induced performance degradation by fine-tuning the adopted models on compressed images, achieving improvements in terms of half total error rate between 5% and 10% for images compressed at the worst JPEG and JPEG AI qualities. We also evaluate the generalizability of the developed solutions by testing them on learning-based codecs not considered during training, to check whether similar PAD-relevant artifacts are introduced by different compressions. Furthermore, we investigate the redundancy within the embeddings generated by the employed detectors, and demonstrated it is possible to significantly compress them while preserving the achievable PAD performance. Overall, the study provides a systematic analysis of iris PAD under compression constraints, offering insights into model adaptation, cross-codec robustness, and representation efficiency in scenarios where visual data coding plays a central role.

Wearable devices have been recently proposed to perform biometric recognition, leveraging on the uniqueness of the collectable physiological traits to generate discriminative identifiers. Most of the studies conducted on this topic have exploited heart-related signals, sensing the cardiac activity either through electrical measurements using electrocardiography, or with optical recordings employing photoplethysmography. In this paper we instead propose a system performing BIOmetric recognition using Wearable Inertial Sensors detecting Heart activity (BIOWISH). In more detail, we investigate the feasibility of exploiting mechanical measurements obtained through seismocardiography and gyrocardiography to verify the identity of a subject. Several feature extractors and classifiers, including deep learning techniques relying on siamese training, are employed to derive distinctive characteristics from the considered signals, so as to differentiate between legitimate users and impostors. A multi-session database, comprising acquisitions taken from subjects performing different activities, is employed to perform experimental tests. The obtained results testify that identifiers derived from measurements of chest vibrations, collected by wearable inertial sensors, could be employed to guarantee high recognition performance, even when considering short-time recordings. Explainability methods have been also employed to derive some insights about the aspects relevant to perform predictions for both people and activity recognition tasks.

Template protection is an issue of paramount importance for the design of secure and privacy-compliant biometric recognition systems. Template unlinkability, together with template irreversibility, is an essential requirement to properly guarantee template protection. In fact, it ensures that templates generated from the same trait, but used in different applications, cannot be linked to the same identity. This paper deals with the design of a system satisfying the unlinkability requirement. The robustness of the proposed solution is evaluated by exploiting methods stemming from the theory of stochastic optimization, as well as by using quantitative measures specifically proposed to characterize the unlinkability of biometric protection schemes. A case study using finger-vein biometrics is considered to test the proposed cryptosystem on non-ideal data. The proposed scheme guarantees 128 bits of security with acceptable false recognition rates in real-life conditions. Moreover, we provide guidelines to determine the parameters of the transformations to be applied to real biometric traits so as to ensure proper recognition, security, and unlinkability performance.

Swapping faces with deep learning technology to generate realistic fake videos/images (a.k.a, deepfakes) has drawn great public con-cerns recently. Numerous approaches have been proposed to iden-tify fake contents; however, less work has been dedicated to pro-tecting the source faces in an active way. In this paper, we stand for a legitimate faceswap service provider and present an approach called Source-ID- Tracker (SIDT), which aims to protect the identity of source faces in deepfakes from malicious uses. As a plug-in, the encoder of SIDT implicitly embeds a source face image into a deep-fake image while ensuring the resultant encoded image is visually indistinguishable from the deepfake image. After sharing through social media, the embedded source face and its identity can still be recovered with a decoder. Experimental results show that the pro-posed model achieves a promising performance, in terms of reconstruction quality and attribution inference accuracy, in revealing the hidden source face.

Biometric characteristics such as fingerprint and face are nowadays employed in several real-world automatic recognition systems. Yet, there is still a notable interest for innovative recognition approaches, possibly involving traits offering properties not available in mainstream solutions. Brain activity is among the novel characteristics recently investigated for biometric recognition, with several studies confirming the existence of subject-specific information within signals acquired through electroencephalography (EEG). Unfortunately, EEG data are typically characterized by a significant intra-class variability, especially when recorded at different times, which makes it hard to design hand-crafted features retaining enough discriminative power to be used in recognition systems. For this reason, deep learning strategies are often employed to extract useful information from the treated data. In this paper, we specifically focus on the exploitation of transfer learning approaches to define reliable biometric templates from EEG traits. Specifically, a time-frequency EEG representation is used as input to networks originally designed for image classification tasks, and here fine-tuned to be employed as feature extractors for biometric recognition purposes. Tests performed in open-set verification scenarios, over a longitudinal database comprising EEG acquisition from 45 subjects spanning a period of more than one year, testify the effectiveness of the proposed approach.

In this letter, we propose a vein-based biometric verification system relying on deep learning. A novel approach consisting of a convolutional neural network (CNN), trained in a supervised manner, cascaded with an auto-encoder, trained in an unsupervised way, is here exploited. In more detail, a novel densely-connected convolutional autoencoder is here used on top of backbone CNNs. This architecture aims at increasing the discriminative capability of the features generated from hand vein patterns. Experimental tests on finger, palm, and dorsal veins show that the proposed approach leads to an improvement of the recognition rates with respect to the use of the sole CNNs for feature extraction. The achieved performance are superior to the current state of the art in vein biometric verification.

Finger-vein-based biometric recognition technology has recently attracted the attention of both academia and industry because of its robustness against presentation attacks and the convenience of the acquisition process. As a matter of fact, some contactless vein-based recognition systems have already been deployed and commercialized. However, they require the users to keep their hands still over the acquisition device for a few seconds to perform recognition. In this study, we release this constraint and allow users to have their finger vein patterns acquired on-the-fly. To accomplish this goal, we introduce an ad-hoc acquisition architecture capable of capturing the finger vein structure using an array of low-cost cameras, and we propose a recognition framework based on the use of convolutional and recurrent neural networks. To test the proposed approach we acquire a finger vein image dataset, in video format at four different exposure times, from 100 subjects. The obtained experimental results show that, even in a very challenging scenario, the proposed system guarantees high performance levels, up to 99.13% recognition accuracy over the collected dataset.

The use of human finger-vein traits for the purpose of automatic user recognition has gained a lot of attention in recent years. Current state-of-the-art techniques can provide relatively good performance, yet they are strongly dependent upon the quality of the analyzed finger-vein images. In this paper, we propose a convolutional-neural-network-based finger-vein identification system and investigate the capabilities of the designed network over four publicly available databases. The main purpose of this paper is to propose a deep-learning method for finger-vein identification, which is able to achieve stable and highly accurate performance when dealing with finger-vein images of different quality. The reported extensive set of experiments show that the accuracy achievable with the proposed approach can go beyond 95% correct identification rate for all the four considered publicly available databases.

This paper deals with electroencephalography (EEG)-based biometric identification, using a motor imagery task, specifically performing imaginary arms and legs movements. Deep learning methods such as convolutional neural network (CNN) is used for automatic discriminative feature extraction and person identification. An extensive set of experimental tests, performed on a large database comprising EEG data collected from 40 subjects over two different sessions taken at a week distance, shows the existence of repeatable discriminative characteristics in individuals' brain signals.

Brain signals have recently attracted the attention of the scientific community as potential biometric identifiers. In more detail, there is a growing interest in evaluating the feasibility of using electroencephalography (EEG) recordings to perform automatic people recognition. In this scenario, the study of the longitudinal behavior of EEG signals, i.e., their permanence across time, is of paramount importance. This paper is the first extensive attempt, in terms of employed elicitation protocols, number of involved subjects, number of acquisition sessions, and covered time span, to evaluate the influence of aging effects on the discriminative capabilities of EEG signals over long-term periods. Specifically, we here report and discuss the results obtained from experimental tests conducted on a database comprising 45 subjects, whose EEG signals have been collected during five to six distinct sessions spanning a total period of three years, using four different elicitation protocols. The longitudinal behavior of EEG discriminative traits is evaluated by means of a statistical-and performance-related analysis, using different EEG features and hidden Markov models as classifiers. A characterization of each considered EEG channel in terms of uniqueness and permanence properties is also performed, with the purpose of ranking their relevance for biometric purposes, thus giving hints to contain their number in practical applications. Moreover, we design some possible countermeasures to mitigate aging effects on recognition performance and evaluate their effectiveness, thus paving the road for the future deployment of real-life cognitive recognition systems relying on brain-based biometric traits.

In this paper, we present a novel biometric cryptosystem obtaining perfect security, that is not leaking any information about the employed secret key from the knowledge of the stored helper data. While similar purposes have already been sought in the literature, the approaches proposed so far have been evaluated in terms of recognition performance under the unrealistic assumption of ideal statistical distributions for the considered biometric data. Conversely, in this paper, we investigate the applicability of the proposed framework to practical scenarios while managing a trade-off between privacy and recognition performance. This goal has been achieved by introducing a class of transformation functions enforcing zero-leakage secrecy, by designing an adaptive strategy for embedding the secret key bits into the selected features, and by developing a system parameters optimization strategy with respect to security, recognition performance, and privacy. Experimental tests conducted on real fingerprint data prove the effectiveness of the proposed scheme.

A blind multi-bit watermarking method, specifically designed for high dynamic range (HDR) images, is introduced in thiswork. The presented embedding strategy takes into account several properties of the human visual system in order to guar-antee proper imperceptibility of the embedded watermarks, as well as robustness against tone-mapping operators (TMOs)employed to derive low dynamic range (LDR) representations of the marked HDR content. Specifically, binary marks areembedded into the discrete wavelet transform (DWT) of the just-noticeable difference domain. Additionally, embeddinglocations are preferably selected in correspondence of the detail parts of an HDR image, determined by exploiting the prop-erties of bilateral filtering. Furthermore, the watermark strength adopted in each DWT subband is modulated depending ona contrast sensitivity function, depending of the subband scale and orientation, to improve the visual quality of the water-marked image. In order to verify the effectiveness of the proposed approach, in terms of both mark imperceptibility androbustness, several experimental tests are carried out on a database comprising 15 HDR images. Specifically, robustness isevaluated by blindly extracting the embedded messages from both HDR marked images and from their LDR counterparts,thus investigating the effects of TMOs on message recovery.

In this letter, we investigate the permanence issue of electroencephalographic (EEG) signals, elicited by visual stimuli, for biometric recognition purposes. Specifically, we evaluate the discriminative capabilities of generic visually-evoked potentials (VEPs) and of visual event-related potentials (ERPs) associated to specific cognitive tasks. Furthermore, we analyze the permanence issue of the considered EEG traits by verifying the stability across time of the achievable recognition rates. Experimental tests performed on a longitudinal database, comprising EEG data taken from 50 subjects during 3 different sessions, give evidence of the presence of repeatable discriminative characteristics in the individuals' EEG activity.

Brain signals have been investigated for more than a century in the medical field. However, despite the broad interest in clinical applications, their use as a biometric identifier has been only recently considered by the scientific community. In this paper, we focus on the permanence across time of brain signals, specifically of electroencephalographic (EEG) signals, issue of paramount importance for the deployment of brain-based biometric recognition systems in real life, not yet fully addressed. In particular, we speculate about the stability of EEG features by analyzing the recognition performance that can be achieved when comparing EEG signals acquired during different sessions. We carry out an extensive set of experimental tests, performed on several EEG-based biometric systems over a large database, comprising three recordings taken from 50 healthy subjects in resting state conditions, acquired in a time span of approximately one month and a half. The results confirm that a significant level of permanence can be guaranteed.

Biometric recognition systems, despite the advantages provided with respect to traditional authentication methods, have some peculiar weaknesses which may allow an attacker being falsely recognized or accessing users' personal data. Among such vulnerabilities, in this paper, we speculate on the hill-climbing attack, i.e., the possibility for an attacker to exploit the scores produced by the matcher with the goal of generating synthetic biometric data, which could allow a false acceptance. More in detail, we focus on multibiometrics systems and investigate about the robustness of different system architectures, both parallel and serial fusion schemes, against the hill-climbing attack. Nonuniform quantization is also evaluated as a possible countermeasure for limiting the effectiveness of the considered attacks in terms of recognition success rate and average number of required attempts without affecting the recognition performance.

Template protection is an issue of paramount importance in the design of biometric recognition systems. In this paper we present a biometric cryptosystem applied to iris biometrics, where template security is guaranteed by means of a framework inspired by the digital modulation paradigm. Specifically, the properties of modulation constellations and turbo codes with soft-decoding are exploited to design a system with high performance in terms of both verification rates and security, even while dealing with a biometrics characterized by a high intra-class variability such as the iris. The effectiveness of the proposed approach is evaluated by performing tests on the Interval subset of the CASIA-IrisV4 database.

Data security and privacy are crucial issues to be addressed for assuring a successful deployment of biometrics-based recognition systems in real life applications. In this paper, a template protection scheme exploiting the properties of universal background models, eigen-user spaces, and the fuzzy commitment cryptographic protocol is presented. A detailed discussion on the security and information leakage of the proposed template protection system is given. The effectiveness of the proposed approach is investigated with application to online signature recognition. The given experimental results, evaluated on the public MCYT signature database, show that the proposed system can guarantee competitive recognition accuracy while providing protection to the employed biometric data.

In this paper we propose a biometric cryptosystem able to provide security and renewability to a function based on- line signature representation. A novel reliable signature traits selection procedure, along with a signature binarization algorithm, are introduced. Experimental results, evaluated on the public MCYT signature database, show that the proposed protected on-line signature recognition system guarantees recognition rates comparable with those of unprotected approaches, and outperforms already proposed protection schemes for signature biometrics.

Recent years have seen the rapid spread of biometric technologies for automatic people recognition. However, security and privacy issues still represent the main obstacles for the deployment of biometric-based authentication systems. In this paper, we propose an approach, which we refer to as BioConvolving, that is able to guarantee security and renewability to biometric templates. Specifically, we introduce a set of noninvertible transformations, which can be applied to any biometrics whose template can be represented by a set of sequences, in order to generate multiple transformed versions of the template. Once the transformation is performed, retrieving the original data from the transformed template is computationally as hard as random guessing. As a proof of concept, the proposed approach is applied to an on-line signature recognition system, where a hidden Markov model-based matching strategy is employed. The performance of a protected on-line signature recognition system employing the proposed BioConvolving approach is evaluated, both in terms of authentication rates and renewability capacity, using the MCYT signature database. The reported extensive set of experiments shows that protected and renewable biometric templates can be properly generated and used for recognition, at the expense of a slight degradation in authentication performance.

In this paper we propose a new method for natural video textures synthesis, using a three-dimensional (3-D) extended self-similar (ESS) fractal model. The autocorrelation functions (ACFs) of original texture increments are estimated, and a synthetic 3-D-ESS process whose increments have the same ACFs of the corresponding given ones is generated using a 3-D incremental Fourier synthesis algorithm. Experimental results for the analysis and synthesis of natural video textures are provided