Morphing Attack Detection (MAD) systems often suffer from performance degradation when deployed in operational environments, such as airports, that differ from the training domain.We propose an adaptive differential MAD framework that continuously refines a pre-trained detector using live bona fide samples acquired at the gate.The system is memoryless, so no samples are stored in memory to mitigate privacy concerns about the collection of personal data.To prevent the loss of discriminative power caused by bona fideonly adaptation, the method generates synthetic morph samples on-the-fly by combining the current operational subject with identities from an external public or synthetic face dataset.The adaptation process further relies on a quality-aware bona fide selection strategy and a controlled balancing mechanism for synthetic morph generation.Experimental results show that the proposed method improves target-domain specialization while maintaining robustness against morph attacks.

Face morphing attacks are widely recognized as one of the most challenging threats to face recognition systems used in electronic identity documents. These attacks exploit a critical vulnerability in passport enrollment procedures adopted by many countries, where the facial image is often acquired without a supervised live capture process. In this paper, we propose a novel face morphing technique based on Arc2Face, an identity-conditioned face foundation model capable of synthesizing photorealistic facial images from compact identity representations. We demonstrate the effectiveness of the proposed approach by comparing the morphing attack potential metric on two large-scale sequestered face morphing attack detection datasets against several state-of-the-art morphing methods, as well as on two novel morphed face datasets derived from FEI and ONOT. Experimental results show that the proposed deep learning-based approach achieves a morphing attack potential comparable to that of landmark-based techniques, which have traditionally been regarded as the most challenging. These findings confirm the ability of the proposed method to effectively preserve and manage identity information during the morph generation process.

Facial recognition is a key biometric technology, especially for using electronic documents in real-world applications. The accuracy of this recognition technology strictly depends on the image quality, i.e. the face appearance in the image included in the document. Then, adherence to ISO/ICAO standards, which contain guidelines to standardize the image quality in official documents, is of paramount importance. However, ensuring compliance is challenging due to high subject variability. Furthermore, controls are often executed manually, making them subjective and time-consuming. Therefore, in this work, we introduce BioGaze, an automated framework for ISO/ICAO compliance verification that combines classical computer vision and deep learning algorithms to perform the checks contained in the latest standard version. The framework is tested on a synthetic dataset, achieving state-of-the-art performance across multiple ISO/ICAO requirements, surpassing public algorithms and commercial SDKs. BioGaze is publicly available to advance automated compliance verification and support standardization efforts1.1https://github.com/MI-BioLab/BioGaze

Ensuring compliance of face images with ISO/ICAO quality standards is essential for boosting the document enrollment process. Indeed, traditional manual checks are slow, subjective, and difficult to scale. Therefore, we propose a system that aims to fully automate compliance verification by directly analyzing the official requirements without relying on predefined hand-crafted features or manual thresholds. Our method combines a Large Language Model, a novel prompt learning procedure, and a contrastive learning framework to evaluate the adherence of a face image to quality requirements. Tested on a recent dataset, our proposed system achieves high accuracy, surpassing existing academic and commercial solutions. By streamlining the implementation and updates to the compliance rules, our approach represents a significant step toward simple, scalable, and regulation-driven image verification. Code and models are publicly available1.

The rapid advancement of generative AI has revolutionized image creation, enabling high-quality synthesis from text prompts while raising critical challenges for media authenticity. We present AI-GenBench, a novel benchmark designed to address the urgent need for robust detection of AI-generated images in real-world scenarios. Unlike existing solutions that evaluate models on static datasets, AI-GenBench introduces a temporal evaluation framework where detection methods are incrementally trained on synthetic images, historically ordered by their generative models, to test their ability to generalize to new generative models, such as the transition from GANs to diffusion models. Our benchmark focuses on high-quality, diverse visual content and overcomes key limitations of current approaches, including arbitrary dataset splits, unfair comparisons, and excessive computational demands. AI-GenBench provides a comprehensive dataset, a standardized evaluation protocol, and accessible tools for both researchers and non-experts (e.g., journalists, fact-checkers), ensuring reproducibility while maintaining practical training requirements. By establishing clear evaluation rules and controlled augmentation strategies, AI-GenBench enables meaningful comparison of detection methods and scalable solutions. Code and data are publicly available to ensure reproducibility and to support the development of robust forensic detectors to keep pace with the rise of new synthetic generators1

Most of today's SLAM approaches learn the map of the environment in the first stage (referred to as mapping) and subsequently use this static map for planning and navigation. This method is suboptimal in dynamic contexts because changes in the environment can result in poor performance of the localization components essential for loop closure detection and relocalization. To address the limitations of the mapping-navigation dualism, continual SLAM has been proposed, which focuses on methods that can continually update the knowledge of the environment and the corresponding map. However, continual SLAM poses challenges, particularly for real-time navigation of large maps, and many of the existing techniques are not yet mature for practical application. In this paper, we present a continual learning approach aimed at accurate and efficient robot localization on large maps, advancing the goal of continual SLAM. Our approach incrementally trains a region prediction neural network to recognize familiar places and preselect a subset of map nodes for localization and map optimization. We integrate this method into RTAB-Map, a well-known graph-based SLAM system, and validate its practical applicability through assessments on several real-world SLAM datasets.

Nowadays, state-of-the-art AI-based generative models represent a viable solution to overcome privacy issues and biases in the collection of datasets containing personal information, such as faces. Following this intuition, in this paper we introduce ONOT11One, No one and One hundred Thousand (L. Pirandello, 1926), a synthetic dataset specifically focused on the generation of high-quality faces in adherence to the requirements of the ISO/IEC 39794–5 standards that, following the guidelines of the International Civil Aviation Organization (ICAO), defines the interchange formats of face images in electronic Machine-Readable Travel Documents (eMRTD). The strictly controlled and varied mugshot images included in ONOT are useful in research fields related to the analysis of face images in eMRTD, such as Morphing Attack Detection and Face Quality Assessment. The dataset is publicly released22https://miatbiolab.csr.unibo.it/icao-synthetic-dataset, in combination with the generation procedure details in order to improve the reproducibility and enable future extensions.

Large-scale face recognition datasets are collected by crawling the Internet and without individuals' consent, raising legal, ethical, and privacy concerns. With the recent advances in generative models, recently several works proposed generating synthetic face recognition datasets to mitigate concerns in web-crawled face recognition datasets. This paper presents the summary of the Synthetic Data for Face Recognition (SDFR) Competition held in conjunction with the 18th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2024) and established to investigate the use of synthetic data for training face recognition models. The SDFR competition was split into two tasks, allowing participants to train face recognition systems using new synthetic datasets and/or existing ones. In the first task, the face recognition backbone was fixed and the dataset size was limited, while the second task provided almost complete freedom on the model backbone, the dataset, and the training pipeline. The submitted models were trained on existing and also new synthetic datasets and used clever methods to improve training with synthetic data. The submissions were evaluated and ranked on a diverse set of seven benchmarking datasets. The paper gives an overview of the submitted face recognition models and reports achieved performance compared to baseline models trained on real and synthetic datasets. Furthermore, the evaluation of submissions is extended to bias assessment across different demography groups. Lastly, an outlook on the current state of the research in training face recognition models using synthetic data is presented, and existing problems as well as potential future directions are also discussed.

In response to the rising threat of the face morphing attack, this paper introduces and explores the potential of Video-based Morphing Attack Detection (V-MAD) systems in real-world operational scenarios. While current morphing attack detection methods primarily focus on a single or a pair of images, V-MAD is based on video sequences, exploiting the video streams acquired by face verification tools available, for instance, at airport gates. We show for the first time the advantages that the availability of multiple probe frames brings to the morphing attack detection task, especially in scenarios where the quality of probe images is varied. Experimental results on a real operational database demonstrate that video sequences represent valuable information for increasing the performance of morphing attack detection systems.

The morphing attack is widely acknowledged as an important security threat to face recognition systems in the context of electronic machine readable travel documents and several possible countermeasures have been recently proposed. Among the existing solutions, differential Morphing Attack Detection (MAD) algorithms, based on the comparison of the document image (possibly morphed) and a trusted live capture, proved to be quite effective and robust in detecting this kind of attack. However, deploying such solutions in a real-world operational scenario requires the capability of dealing with images of variable quality in terms of illumination, pose, focus, etc. This paper analyzes the impact of face image quality on MAD performance through an extensive image quality assessment, carried out on a large and realistic operational dataset using different state-of-the-art algorithms, thus providing useful insights for the development of more robust MAD systems.

Through the Face Morphing attack is possible to use the same legal document by two different people, destroying the unique biometric link between the document and its owner. In other words, a morphed face image has the potential to bypass face verification-based security controls, then representing a severe security threat. Unfortunately, the lack of public, extensive and varied training datasets severely hampers the development of effective and robust Morphing Attack Detection (MAD) models, key tools in contrasting the Face Morphing attack since able to automatically detect the presence of morphing images. Indeed, privacy regulations limit the possibility of acquiring, storing, and transferring MAD-related data that contain personal information, such as faces. Therefore, in this paper, we investigate the use of Federated Learning to train a MAD model on local training samples across multiple sites, eliminating the need for a single centralized training dataset, as common in Machine Learning, and then overcoming privacy limitations. Experimental results suggest that FL is a viable solution that will need to be considered in future research works in MAD.

A better understanding of the emergent computation and problem-solving capabilities of recent large language models is of paramount importance to further improve them and broaden their applicability. This work investigates how a language model, trained to predict the next token, can perform arithmetic computations generalizing beyond training data. Binary addition and multiplication constitute a good testbed for this purpose, since they require a very small vocabulary and exhibit relevant input/output discontinuities making smooth input interpolation ineffective for novel data. We successfully trained a light language model to learn these tasks and ran a number of experiments to investigate the extrapolation capabilities and internal information processing. Our findings support the hypothesis that the language model works as an Encoding-Regression-Decoding machine where the computation takes place in the value space once the input token representation is mapped to an appropriate internal representation.

Scenarios in which restrictions in data transfer and storage limit the possibility to compose a single dataset – also exploiting different data sources – to perform a batch-based training procedure, make the development of robust models particularly challenging. We hypothesize that the recent Continual Learning (CL) paradigm may represent an effective solution to enable incremental training, even through multiple sites. Indeed, a basic assumption of CL is that once a model has been trained, old data can no longer be used in successive training iterations and in principle can be deleted. Therefore, in this paper, we investigate the performance of different Continual Learning methods in this scenario, simulating a learning model that is updated every time a new chunk of data, even of variable size, is available. Experimental results reveal that a particular CL method, namely Learning without Forgetting (LwF), is one of the best-performing algorithms. Then, we investigate its usage and parametrization in Morphing Attack Detection and Object Classification tasks, specifically with respect to the amount of new training data that became available.

Recently, the Face Morphing Attack has emerged as a serious and concrete security threat, and then several Morphing Attack Detectors (MADs) have been proposed in the literature. Unfortunately, most MAD approaches, especially if based on single input images, are not yet mature for real-world deployment, mainly due to low accuracy and limited generalization capabilities with data distribution different from those used for training. While better models and techniques will certainly contribute to advancing the state-of-the-art, one of the main obstacles is on the data side: indeed, training data available to research groups are often limited in size and variety, and privacy issues restrict the possibility of data release and exchange. The proposed approach envisages the incremental training of MADs on a collection of datasets even owned by different research groups, adopting a learning strategy that involves model transfer as an alternative to data sharing. Specifically, in this paper, we propose and publicly release a framework to support the adoption of Continual Learning strategies, enabling an efficient incremental training for MADs on new data that progressively become available at different places or times. Different learning strategies are analyzed and compared in terms of the effectiveness and stability of the results achieved.

On-device training for personalized learning is a challenging research problem.Being able to quickly adapt deep prediction models at the edge is necessary to better suit personal user needs.However, adaptation on the edge poses some questions on both the efficiency and sustainability of the learning process and on the ability to work under shifting data distributions.Indeed, naively fine-tuning a prediction model only on the newly available data results in catastrophic forgetting, a sudden erasure of previously acquired knowledge.In this paper, we detail the implementation and deployment of a hybrid continual learning strategy (AR1*) on a native Android application for real-time on-device personalization without forgetting.Our benchmark, based on an extension of the CORe50 dataset, shows the efficiency and effectiveness of our solution.23

Morphing attacks have posed a severe threat to Face Recognition System (FRS). Despite the number of advancements reported in recent works, we note serious open issues such as independent benchmarking, generalizability challenges and considerations to age, gender, ethnicity that are inadequately addressed. Morphing Attack Detection (MAD) algorithms often are prone to generalization challenges as they are database dependent. The existing databases, mostly of semi-public nature, lack in diversity in terms of ethnicity, various morphing process and post-processing pipelines. Further, they do not reflect a realistic operational scenario for Automated Border Control (ABC) and do not provide a basis to test MAD on unseen data, in order to benchmark the robustness of algorithms. In this work, we present a new sequestered dataset for facilitating the advancements of MAD where the algorithms can be tested on unseen data in an effort to better generalize. The newly constructed dataset consists of facial images from 150 subjects from various ethnicities, age-groups and both genders. In order to challenge the existing MAD algorithms, the morphed images are with careful subject pre-selection created from the contributing images, and further post-processed to remove morphing artifacts. The images are also printed and scanned to remove all digital cues and to simulate a realistic challenge for MAD algorithms. Further, we present a new online evaluation platform to test algorithms on sequestered data. With the platform we can benchmark the morph detection performance and study the generalization ability. This work also presents a detailed analysis on various subsets of sequestered data and outlines open challenges for future directions in MAD research.

In this paper we present a novel defogging technique, named CurL-Defog, aimed at minimizing the creation of unwanted artifacts during the defogging process. The majority of learning based defogging approaches rely on paired data (i.e., the same images with and without fog), where fog is artificially added to clear images: this often provides good results on mildly fogged images but does not generalize well to real difficult cases. On the other hand, the models trained with real unpaired data (e.g. CycleGAN) can provide visually impressive results but they often produce unwanted artifacts. In this paper we propose a curriculum learning strategy coupled with an enhanced CycleGAN model in order to reduce the number of produced artifacts, while maintaining state-of-the-art performance in terms of contrast enhancement and image reconstruction. We also introduce a new metric, called HArD (Hazy Artifact Detector) to numerically quantify the amount of artifacts in the defogged images, thus avoiding the tedious and subjective manual inspection of the results. The proposed approach compares favorably with state-of-the-art techniques on both real and synthetic datasets.

Training deep neural networks at the edge on light computational devices, embedded systems and robotic platforms is nowadays very challenging. Continual learning techniques, where complex models are incrementally trained on small batches of new data, can make the learning problem tractable even for CPU-only embedded devices enabling remarkable levels of adaptiveness and autonomy. However, a number of practical problems need to be solved: catastrophic forgetting before anything else. In this paper we introduce an original technique named "Latent Replay" where, instead of storing a portion of past data in the input space, we store activations volumes at some intermediate layer. This can significantly reduce the computation and storage required by native rehearsal. To keep the representation stable and the stored activations valid we propose to slow-down learning at all the layers below the latent replay one, leaving the layers above free to learn at full pace. In our experiments we show that Latent Replay, combined with existing continual learning techniques, achieves state-of-the-art performance on complex video benchmarks such as CORe50 NICv2 (with nearly 400 small and highly non-i.i.d. batches) and OpenLORIS. Finally, we demonstrate the feasibility of nearly real-time continual learning on the edge through the deployment of the proposed technique on a smartphone device.

It was recently shown that architectural, regularization and rehearsal strategies can be used to train deep models sequentially on a number of disjoint tasks without forgetting previously acquired knowledge. However, these strategies are still unsatisfactory if the tasks are not disjoint but constitute a single incremental task (e.g., class-incremental learning). In this paper we point out the differences between multi-task and single-incremental-task scenarios and show that well-known approaches such as LWF, EWC and SI are not ideal for incremental task scenarios. A new approach, denoted as AR1, combining architectural and regularization strategies is then specifically proposed. AR1 overhead (in terms of memory and computation) is very small thus making it suitable for online learning. When tested on CORe50 and iCIFAR-100, AR1 outperformed existing regularization strategies by a good margin.

The morphing attack proved to be a serious threat for modern automated border control systems where face recognition is used to link the identity of a passenger to his/her e-document. In this paper, we show that by exploiting the live face image acquired at the gate, the morphed face image stored in the document can be reverted (or demorphed) enough to reveal the identity of the legitimate document owner, thus allowing the system to issue a warning. A number of practical experiments on two data sets proves the efficacy of our approach.

A double-identity fingerprint is a fake fingerprint created by combining features from two different fingers, so that it has a high chance to be falsely matched with fingerprints from both fingers. This paper studies the feasibility of creating double-identity fingerprints by proposing two possible techniques and evaluating to what extent they may be used to fool the state-of-the-art fingerprint recognition systems. The results of systematic experiments suggest that existing algorithms are highly vulnerable to this specific attack (about 90% chance of success at FAR = 0.1%) and that the fingerprint patterns generated might be realistic enough to fool human examiners.

Recent works demonstrated the usefulness of temporal coherence to regularize supervised training or to learn invariant features with deep architectures. In particular, enforcing a smooth output change while presenting temporally-closed frames from video sequences, proved to be an effective strategy. In this paper we prove the efficacy of temporal coherence for semi-supervised incremental tuning. We show that a deep architecture, just mildly trained in a supervised manner, can progressively improve its classification accuracy, if exposed to video sequences of unlabeled data. The extent to which, in some cases, a semi-supervised tuning allows to improve classification accuracy (approaching the supervised one) is somewhat surprising. A number of control experiments pointed out the fundamental role of temporal coherence.

Once upon a time there was a criminal; he was reading his e-mail when a banner caught his attention: low cost flights for the destination of his dreams! He had already started to book the trip when suddenly realized that, being wanted by the police, he could not use his passport without being arrested. What to do? He could not miss that opportunity, so he called a good friend and they started to think for a possible solution. Do you want to know if they succeeded? Read the rest of the paper and find it out.

Hierarchical temporal memory (HTM) is a biologically inspired framework that can be used to learn invariant representations of patterns in a wide range of applications. Classical HTM learning is mainly unsupervised, and once training is completed, the network structure is frozen, thus making further training (i.e., incremental learning) quite critical. In this letter, we develop a novel technique for HTM (incremental) supervised learning based on gradient descent error minimization. We prove that error backpropagation can be naturally and elegantly implemented through native HTM message passing based on belief propagation. Our experimental results demonstrate that a two-stage training approach composed of unsupervised pretraining and supervised refinement is very effective (both accurate and efficient). This is in line with recent findings on other deep architectures.

This paper introduces a novel method for the generation of high-resolution synthetic hand-print images. Specific traits, such as fingerprint, palmprint, and hand-shape, are synthesized to obtain a whole hand-print. Each trait is generated by a methodology that mimics the nature of the corresponding biometric data and their main degrees of freedom. The biometric traits are then integrated into a single high-resolution realistic image. A quantitative validation of the obtained patterns is carried out in the context of minutiae matching by comparing genuine and impostor distributions between synthetic and real hand-prints. The proposed approach also proved to be useful for algorithm training/optimization.

Face images to be included into machine readable travel documents have to fulfill quality requirements defined by international ISO standards. The concept of quality in this context extends the common idea of image quality: usually a bad quality image presents visual defects such as blurring or noise while, according to ISO/ICAO standard, other factors could make a given sample a poor quality image (e.g., presence of dark glasses or mouth open). The verification of face image conformance to ISO/ICAO standards is carried out mostly by humans today, through visual inspection, since a totally automatic evaluation is still not satisfactory. The objective of this work is to present the BioLab-ICAO framework, an evaluation benchmark which will be made available to the scientific community, designed to encourage the research on this topic; it consists of a large ground truth database, a well-defined testing protocol, and baseline algorithms for image compliance verification.

Although several fingerprint template protection methods have been proposed in the literature, the problem is still unsolved, since enforcing nonreversibility tends to produce an excessive drop in accuracy. Furthermore, unlike fingerprint verification, whose performance is assessed today with public benchmarks and protocols, performance of template protection approaches is often evaluated in heterogeneous scenarios, thus making it very difficult to compare existing techniques. In this paper, we propose a novel protection technique for Minutia Cylinder-Code (MCC), which is a well-known local minutiae representation. A sophisticate algorithm is designed to reverse MCC (i.e., recovering original minutiae positions and angles). Systematic experimentations show that the new approach compares favorably with state-of-the-art methods in terms of accuracy and, at the same time, provides a good protection of minutiae information and is robust against masquerade attacks.

This paper summarizes the results of the fingerprint verification competition organized in conjunction with IJCB 2011. The competition focused on benchmarks covering both proprietary encoding and ISO template format. Considering the benchmarks difficulty, some of the algorithms submitted achieved very good accuracy: a 0.7% EER and a 1.1% EER were obtained on two challenging benchmarks, using proprietary and ISO template formats, respectively. Based on the participant self-description of the best performing algorithms we tried to figure out the most promising building-block technologies.

This paper proposes a new hash-based indexing method to speed up fingerprint identification in large databases. A Locality-Sensitive Hashing (LSH) scheme has been designed relying on Minutiae Cylinder-Code (MCC), which proved to be very effective in mapping a minutiae-based representation (position/ angle only) into a set of fixed-length transformation-invariant binary vectors. A novel search algorithm has been designed thanks to the derivation of a numerical approximation for the similarity between MCC vectors. Extensive experimentations have been carried out to compare the proposed approach against 15 existing methods over all the benchmarks typically used for fingerprint indexing. In spite of the smaller set of features used (top performing methods usually combine more features), the new approach outperforms existing ones in almost all of the cases.

Computation of local orientations is a primary step in fingerprint recognition. A large number of approaches have been proposed in the literature, but no systematic quantitative evaluations have been done yet. We implemented and tested several well know methods and a plethora of their variants over a novel, specifically designed, benchmark, made available in the FVC-onGoing framework. We proved that parameter optimizations, pre- and post-processing stages can markedly improve accuracy of the baseline methods on bad quality fingerprints. Finally, in this paper we propose a novel adaptive method which selectively exploits accuracy of local-based analysis and learning-based global methods, thus achieving the overall best performance on a challenging dataset.

This paper describes an improved version of the MCC fingerprint matching approach. An in-depth error analysis allowed us to point out the weakest points of the original MCC and to design: i) a more effective minutiae pair selection and ii) a more distortion-tolerant relaxation. The parameters of the new version have been tuned over a new larger dataset and the final algorithm has been evaluated on FVC-onGoing. The results show that MCC compares favorably with some of the most accurate commercial algorithms published in FVC-onGoing.

The computation of local orientations is a fundamental step in fingerprint recognition. Although a large number of approaches have been proposed in the literature, no systematic quantitative evaluations have been done yet, mainly due to the lack of proper datasets with associated ground truth information. In this paper we propose a new benchmark (which includes two datasets and an accuracy metric) and report preliminary results obtained by testing four well-known local orientation extraction algorithms.

In this paper, we introduce the Minutia Cylinder-Code (MCC): a novel representation based on 3D data structures (called cylinders), built from minutiae distances and angles. The cylinders can be created starting from a subset of the mandatory features (minutiae position and direction) defined by standards like ISO/IEC 19794-2 (2005). Thanks to the cylinder invariance, fixed-length, and bit-oriented coding, some simple but very effective metrics can be defined to compute local similarities and to consolidate them into a global score. Extensive experiments over FVC2006 databases prove the superiority of MCC with respect to three well-known techniques and demonstrate the feasibility of obtaining a very effective (and interoperable) fingerprint recognition implementation for light architectures.

This work focuses on performance assessment of software applications designed to evaluate the compliance of a face image to the ISO/ICAO standards for machine readable travel documents. In this paper we describe the new large database (of compliant and non-compliant images) we gathered, the associated testing protocol and the preliminary results measured on some existing algorithms.

Fingerprint singularities play an important role in several fingerprint recognition and classification systems. Although some general relationships and constraints about the location of singularities in the different fingerprint classes are well known, to the best of our knowledge no statistical models have been developed until now. This paper studies the spatial distributions of singularity locations in nature and derives, from a representative dataset of labelled samples, the probability density functions of the four main fingerprint classes. The results obtained can be directly exploited to improve the accuracy of many techniques relying on the position of singularities, as confirmed by the results of two experiments on fingerprint classification and synthesis.

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.

This paper addresses the problem of evaluating the ldquooperational qualityrdquo of fingerprint scanners, that is, the ability of acquiring images that maximize the accuracy of automated fingerprint recognition. The quality parameters commonly used to quantify the fidelity of a scanner in sensing the input pattern have been analyzed and a large experimentation has been carried out to understand their effects on fingerprint recognition accuracy. The experimental results show that some parameters have a strong impact, while others appear to be less relevant.

A minutiae-based template is a very compact representation of a fingerprint image and for a long time it has been assumed that it did not contain enough information to allow the reconstruction of the original fingerprint. This work proposes a novel approach to reconstruct fingerprint images from standard templates and investigates to what extent the reconstructed images are similar to the original ones (i.e., those the templates were extracted from). The efficacy of the reconstruction technique has been assessed by estimating the success chances of a masquerade attack against nine different fingerprint recognition algorithms. The experimental results show that the reconstructed images are very realistic and that, although it is unlikely they can fool a human expert, there is a high chance to deceive state-of-the-art commercial fingerprint recognition systems.

The guest editors provide an overview of the articles selected for this special issue. The issue's goal is to document the current state-of-the-art, acknowledge the latest breakthroughs achieved by scientists working in the area of biometric recognition, and identify future promising research areas. It is thought the selection of papers discussed should give readers a good idea of where researchers have been focusing, both on long- studied problems still needing more work and on newer challenges. A fundamental of the field of biometrics is an ever-increasing need for better recognition and stronger security. But, as public and commercial biometric deployments increase in number, there is also more need to understand privacy issues and to provide greater ease-of-use. The volume and quality of papers in this special issue indicate that much progress has been made in many aspects of the biometrics field and that there are challenging and promising future directions still to follow.

For a long time it has been assumed that a minutiae-based fingerprint template did not contain enough information to allow reconstructing the original fingerprint image. This work proposes an approach to reconstruct fingerprint images from ISO standard templates and investigates to what extent the reconstructed fingerprints are similar to the original ones. The results show that the reconstructed images have a very good quality and may be used to attack existing fingerprint recognition systems. Systematic studies to understand the success chances of such attacks are currently being carried out

This paper is concerned with the performance evaluation of fingerprint verification systems. After an initial classification of biometric testing initiatives, we explore both the theoretical and practical issues related to performance evaluation by presenting the outcome of the recent Fingerprint Verification Competition (FVC2004). FVC2004 was organized by the authors of this work for the purpose of assessing the state-of-the-art in this challenging pattern recognition application and making available a new common benchmark for an unambiguous comparison of fingerprint-based biometric systems. FVC2004 is an independent, strongly supervised evaluation performed at the evaluators' site on evaluators' hardware. This allowed the test to be completely controlled and the computation times of different algorithms to be fairly compared. The experience and feedback received from previous, similar competitions (FVC2000 and FVC2002) allowed us to improve the organization and methodology of FVC2004 and to capture the attention of a significantly higher number of academic and commercial organizations (67 algorithms were submitted for FVC2004). A new, "Light" competition category was included to estimate the loss of matching performance caused by imposing computational constraints. This paper discusses data collection and testing protocols, and includes a detailed analysis of the results. We introduce a simple but effective method for comparing algorithms at the score level, allowing us to isolate difficult cases (images) and to study error correlations and algorithm "fusion." The huge amount of information obtained, including a structured classification of the submitted algorithms on the basis of their features, makes it possible to better understand how current fingerprint recognition systems work and to delineate useful research directions for the future.

Attacking fingerprint-based biometric systems by presenting fake fingers at the sensor could be a serious threat for unattended applications. This work introduces a new approach for discriminating fake fingers from real ones, based on the analysis of skin distortion. The user is required to move the finger while pressing it against the scanner surface, thus deliberately exaggerating the skin distortion. Novel techniques for extracting, encoding and comparing skin distortion information are formally defined and systematically evaluated over a test set of real and fake fingers. The proposed approach is privacy friendly and does not require additional expensive hardware besides a fingerprint scanner capable of capturing and delivering frames at proper rate. The experimental results indicate the new approach to be a very promising technique for making fingerprint recognition systems more robust against fake-finger-based spoofing attempts.

This paper introduces a new noise model for the generation of synthetic fingerprints. Starting from several visual observations about noise in real fingerprints, the limits of the uniform noise generation adopted in our previous work are pointed out and a new model able to create realistic coherent noise is proposed. The results are very promising: the new images generated are definitely more realistic than those obtained with the original method.

This paper introduces a new face recognition approach that allows face variations (produced by aging and other appearance changes) to be dealt with. During the initial learning, a set of MKL subspaces is created for each individual, starting from the feature vectors extracted through a bank of Gabor filters. Then, during the normal system operation, an incremental updating technique can be applied to adjust the subspaces without recalculating the face models from scratch; this makes the method able to cope with gradual changes that occur over time. The results of the experimentation performed on three face databases prove the advantages of the proposed approach with respect to other well-known techniques; in particular, our method achieves better accuracy and higher robustness against face variations.

Reliable and accurate fingerprint recognition is a challenging pattern recognition problem, requiring algorithms robust in many contexts. FVC2000 competition attempted to establish the first common benchmark, allowing companies and academic institutions to unambiguously compare performance and track improvements in their fingerprint recognition algorithms. Three databases were created using different state-of-the-art sensors and a fourth database was artificially generated; 11 algorithms were extensively tested on the four data sets. We believe that FVC2000 protocol, databases, and results will be useful to all practitioners in the field not only as a benchmark for improving methods, but also for enabling an unbiased evaluation of algorithms.

This work introduces the multispace Karhunen-Loeve (MKL) as a new approach to unsupervised dimensionality reduction for pattern representation and classification. The training set is automatically partitioned into disjoint subsets, according to an optimality criterion; each subset then determines a different KL subspace which is specialized in representing a particular group of patterns. The extension of the classical KL operators and the definition of ad hoc distances allow MKL to be effectively used where KL is commonly employed. The limits of the standard KL transform are pointed out, in particular, MKL is shown to outperform KL when the data distribution is far from a multidimensional Gaussian and to better cope with large sets of patterns, which could cause a severe performance drop in KL.

Introduces a method for the generation of synthetic fingerprint images. Gabor-like space-variant filters are used for iteratively expanding an initially empty image containing just one or a few seeds. A directional image model, whose inputs are the number and location of the fingerprint cores and deltas, is used for tuning the filters according to the underlying ridge orientation. Very realistic fingerprint images are obtained after the final noising-and-rendering stage.

In this work, we introduce a new approach to automatic fingerprint classification. The directional image is partitioned into "homogeneous" connected regions according to the fingerprint topology, thus giving a synthetic representation which can be exploited as a basis for the classification. A set of dynamic masks, together with an optimization criterion, are used to guide the partitioning. The adaptation of the masks produces a numerical vector representing each fingerprint as a multidimensional point, which can be conceived as a continuous classification. Different search strategies are discussed to efficiently retrieve fingerprints both with continuous and exclusive classification. Experimental results have been given for the most commonly used fingerprint databases and the new method has been compared with other approaches known in the literature: As to fingerprint retrieval based on continuous classification, our method gives the best performance and exhibits a very high robustness.

This paper introduces a new efficient method for estimating the local ridge-line density in digital images. A mathematical characterization of the local frequency of sinusoidal signals is given, and a 2D-model is developed in order to approximate the ridge-line patterns. Experimental results obtained through a discrete implementation of the method are presented both in terms of accuracy and efficiency.

Minutiae correspond essentially to the terminations and bifurcations of fingerprint patterns. Since the quality of fingerprint images is often low, automatic minutiae detection is a very difficult task and the extraction algorithms produce a large number of false alarms. We present an approach to minutiae filtering based on a neural network. The minutiae neighborhoods extracted by the algorithm presented by us (1997) are normalized with respect to rotation and scale, and their dimensionality is reduced via a KL transform. A neural classifier, whose topology has been designed to exploit the minutiae duality, is employed to perform the neighborhoods classification. The filtering proposed, as confirmed by simulations, allows a significant improvement in the overall performance to be achieved.

We address the problem of performance evaluation in biometric verification systems. By formulating the optimum Bayesian decision criterion for a verification system and by assuming the data distributions to be multinormals, we derive two statistical expressions for calculating theoretically the false acceptance and false rejection rates. Generally, the adoption of a Bayesian parametric model does not allow for obtaining explicit expressions for the calculation of the system errors. As far as biometric verification systems are concerned, some hypotheses can be reasonably adopted, thus allowing simple and affordable expressions to be derived. By using two verification system prototypes. Based on hand shape and human face, respectively, we show our results are well founded.

Most automatic systems for fingerprint comparison are based on minutiae matching. Minutiae are essentially terminations and bifurcations of the ridge lines that constitute a fingerprint pattern. Automatic minutiae detection is an extremely critical process, especially in low-quality fingerprints where noise and contrast deficiency can originate pixel configurations similar to minutiae or hide real minutiae. Several approaches have been proposed in the literature; although rather different from each other, all these methods transform fingerprint images into binary images. In this work we propose an original technique, based on ridge line following, where the minutiae are extracted directly from gray scale images. The results achieved are compared with those obtained through some methods based on image binarization. In spite of a greater conceptual complexity, the method proposed performs better both in terms of efficiency and robustness.

A new structural approach to the fingerprint classification problem is presented. The fingerprint directional image is segmented in regions by minimizing the variance of the element directions within the regions. From the segmentation of the directional image a relational graph compactly summarising the macro-structure of the fingerprint is derived. Inexact graph matching techniques can be adopted to compare the obtained graph with the model graphs in order to classify the fingerprint, according to a given classification scheme.

The problem of organizing and exploiting spatial knowledge for navigation is an important issue in the field of autonomous mobile systems. In particular, partitioning the environment map into connected clusters allows for significant topological features to be captured and enables decomposition of path-planning tasks through a divide-and-conquer policy. Clustering by discovery is a procedure for identifying clusters in a map being learned by exploration as the agent moves within the environment, and yields a valid clustering of the available knowledge at each exploration step. In this work, we define a fitness measure for clustering and propose two incremental heuristic algorithms to maximize it. Both algorithms determine clusters dynamically according to a set of topological and metric criteria. The first one is aimed at locally minimizing a measure of "scattering" of the entities belonging to clusters, and partially rearranges the existing clusters at each exploration step. The second estimates the positions and dimensions of clusters according to a global map of density. The two algorithms are compared in terms of optimality, efficiency, robustness, and stability.