Papers by Gian Luca Marcialis
Current methods for automatic template update are aimed at capturing large intra-class variations... more Current methods for automatic template update are aimed at capturing large intra-class variations of input data and at the same time restricting the probability of impostor’s introduction in client’s galleries. These automatic methods avoid the costs of supervised update methods, which are due to repeated enrollment sessions and manual assignment of identity labels. Most of state-of-the-art template update approaches add input patterns to the claimed identity’s gallery on the basis of their matching score with the existing templates, which must be above a very high “updating” threshold. However, regardless of the value of such updating threshold, update errors do exist and impact strongly on the effectiveness of update procedures. The introduction of impostors into the galleries may degrade the performance quickly. This effect has not been studied in the literature so far. Therefore, a first experimental investigation is the goal of this paper, with a case study on a face verification system.
Although fingerprint verification systems reached a high degree of accuracy, it has been recently... more Although fingerprint verification systems reached a high degree of accuracy, it has been recently shown that they can be circumvented by “fake fingers”, namely, fingerprint images coming from stamps reproducing an user fingerprint, which is processed as an “alive” one. Several methods have been proposed for facing with this problem, but the issue is far from a final solution. Since the problem is relevant both for the academic and the industrial communities, in this paper, we present a critical review of current approaches to fingerprint vitality detection in order to analyze the state–of–the art and the related open issues.
We present new fingerprint classification algorithms based on two machine learning approaches: su... more We present new fingerprint classification algorithms based on two machine learning approaches: support vector machines (SVMs), and recursive neural networks (RNNs). RNNs are trained on a structured representation of the fingerprint image. They are also used to extract a set of distributed features which can be integrated in the SVMs. SVMs are combined with a new error correcting code scheme which, unlike previous systems, can also exploit information contained in ambiguous fingerprint images. Experimental results indicate the benefit of integrating global and structured representations and suggest that SVMs are a promising approach for fingerprint classification.
Journal of Electronic Imaging, Jan 1, 2008
Although many image quality measures have been proposed for fingerprints, few works have taken in... more Although many image quality measures have been proposed for fingerprints, few works have taken into account how differences among capture devices impact on the image quality. In this paper, several representative measures for assessing the quality of fingerprint images are compared using an optical and a capacitive sensor. We implement and test a representative set of measures that rely on different fingerprint image features for quality assessment. The capability to discriminate between images of different quality and its relationship with the verification performance is studied.
International Journal of Image and Graphics, Jan 1, 2008
Although several image quality measures have been proposed for fingerprints, no work has taken in... more Although several image quality measures have been proposed for fingerprints, no work has taken into account the differences among capture devices, and how these differences impact on the image quality. In this paper, several representative measures for assessing the quality fingerprint images are compared using an optical and a capacitive sensor. The capability to discriminate between images of different quality and its relationship with the verification performance is studied. We report differences depending on the sensor, and interesting relationships between sensor technology and features used for quality assessment are also pointed out.
In this paper, an experimental comparison among three structural approaches to fingerprint classi... more In this paper, an experimental comparison among three structural approaches to fingerprint classification is reported. Main pros and cons of such approaches are investigated by experiments and discussed. Moreover, the effectiveness of their measurement-level fusion is analysed. Finally, a comparison among the investigated structural approaches and the well-known statistical approach based on “FingerCodes” is reported.
Template representativeness is a fundamental problem in a biometric recognition system. The perfo... more Template representativeness is a fundamental problem in a biometric recognition system. The performance of the system degrades if the enrolled templates are un-representative of the substantial intra-class variations encountered in the input biometric samples. Recently, several template updates methods based on supervised and semi-supervised learning have been proposed in the literature with an aim to update the enrolled templates to the intra-class variations of the input data. However, the state of art related to template update is still in its infancy. This paper presents a critical review of the current approaches to template updating in order to analyze the state of the art in terms of advancement reached and open issues remain.
Journal of Visual Languages and Computing, Jan 1, 2009
Soft biometrics have been recently proposed for improving the verification performance of biometr... more Soft biometrics have been recently proposed for improving the verification performance of biometric recognition systems. Examples of soft biometrics are skin, eyes, hair colour, height, and ethnicity. Some of them are often cheaper than “hard”, standard biometrics (e.g., face and fingerprints) to extract. They exhibit a low discriminant power for recognizing persons, but can add some evidences about the personal identity, and can be useful for a particular set of users. In particular, it is possible to argue that users with a certain high discriminant soft biometric can be better recognized. Identifying such users could be useful in exploiting soft biometrics at the best, as deriving an appropriate methodology for embedding soft-biometric information into the score computed by the main biometric.In this paper, we propose a group-specific algorithm to exploit soft-biometric information in a biometric verification system. Our proposal is exemplified using hair colour and ethnicity as soft biometrics and face as biometric. Hair colour and information about ethnicity can be easily extracted from face images, and used only for a small number of users with highly discriminant hair colour or ethnicity. We show by experiments that for those users, hair colour or ethnicity strongly contributes to reduce the false rejection rate without a significant impact on the false acceptance rate, whilst the performance does not change for other users.
In this paper, a neural fusion rule for fingerprint verification is presented. The person to be i... more In this paper, a neural fusion rule for fingerprint verification is presented. The person to be identified submits to the system her/his fingerprint and her/his identity. Multiple fingerprint matchers provide a set of verification scores, that are then fused by a perceptronbased method. The weights of such perceptron are explicitly optimised to increase the separation between genuine users and impostors (i.e., unknown users). To this end, the perceptron learning algorithm was modified. Reported experiments show that such modified perceptron allows improving the performances and the robustness of the best individual fingerprint matcher, and outperforming some simple fusion rules.
The user has requested enhancement of the downloaded file. All in-text references underlined in b... more The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document and are linked to publications on ResearchGate, letting you access and read them immediately.
Classification is an important step towards fingerprint recognition. In the classification stage,... more Classification is an important step towards fingerprint recognition. In the classification stage, fingerprints are usually associated to one of the five classes “A”, “L”, “R”, “T”, “W”. The aim is to reduce the number of comparisons that are necessary for recognition. Many approaches to fingerprint classification have been proposed so far, but very few works investigated the potentialities of combining statistical and structural algorithms. In this paper, an approach to fusion of statistical and structural fingerprint classifiers is presented and experiments that show the potentialities of such fusion are reported.
Performances of biometric recognition systems can degrade quickly when the input biometric traits... more Performances of biometric recognition systems can degrade quickly when the input biometric traits exhibit substantial variations compared to the templates collected during the enrolment stage of users. This issue can be addressed using template update methods. In this paper, a novel template update method based on the concept of biometric co-training is presented. In multimodal biometric systems, this method allows co-updating the template galleries of different biometrics, realizing a co-training process of biometric experts which allows updating templates more quickly and effectively. Reported results provide a first experimental evidence of the effectiveness of the proposed template update method.
Pattern Analysis and Applications, Jan 1, 2004
Although many algorithms have been proposed, face recognition and verification systems can guaran... more Although many algorithms have been proposed, face recognition and verification systems can guarantee a good level of performances only for controlled environments. In order to improve the performance and robustness of face recognition and verification systems, multi-modal and mono-modal systems based on the fusion of multiple recognisers using different or similar biometrics have been proposed, especially for verification purposes. In this paper, a recognition and verification system based on the combination of two well-known appearance-based representations of the face, namely, principal component analysis (PCA) and linear discriminant analysis (LDA), is proposed. Both PCA and LDA are used as feature extractors from frontal view images. The benefits of such a fusion are shown for different environmental conditions, namely, ideal conditions, characterised by a very limited variability of environmental parameters, and real conditions with a large variability of lighting, scale and facial expression.
Pattern Recognition Letters, Jan 1, 2005
In this paper, a perceptron-based algorithm for fusion of multiple fingerprint matchers is presen... more In this paper, a perceptron-based algorithm for fusion of multiple fingerprint matchers is presented. The person to be identified submits to the personal authentication system her/his fingerprint and claimed identity. Multiple fingerprint matchers provide a set of verification scores, that are then fused by a single-layer perceptron with class-separation loss function. Weights of such perceptron are explicitly optimised to increase the separation between genuine users and impostors (i.e., unauthorized users). Reported experiments show that such modified perceptron allows improving the performances and the robustness of the best individual fingerprint matcher, and outperforming some commonly used fusion rules.
Although many approaches for face recognition have been proposed in the last years, none of them ... more Although many approaches for face recognition have been proposed in the last years, none of them can overcome the main problem of this kind of biometrics: the huge variability of many environmental parameters (lighting, pose, scale). Hence, face recognition systems can achieve good results at the expense of robustness. In this work we describe a methodology for improving the robustness of a face recognition system based on the "fusion" of two well-known statistical representations of a face: PCA and LDA. Experimental results that confirm the benefits of fusing PCA and LDA are reported.
This paper investigates the advantages of the combination of flat and structural approaches for f... more This paper investigates the advantages of the combination of flat and structural approaches for fingerprint classification. A novel structural classification method is described and compared with the “multichannel” flat method recently proposed by Jain et al. [1]. Performances and complementarity of the two methods are evaluated using NIST-4 Database. A simple approach based on the concept of “metaclassification” is proposed for the combination of the two fingerprint classification methods. Reported results point out the potential advantages of the combination of flat and structural fingerprint-classification approaches. In particular, such results show that the exploitation of structural information allows increasing classification performances.
Pattern Recognition, Jan 1, 2005
Dynamic classifier selection (DCS) plays a strategic role in the field of multiple classifier sys... more Dynamic classifier selection (DCS) plays a strategic role in the field of multiple classifier systems (MCS). This paper proposes a study on the performances of DCS by Local Accuracy estimation (DCS-LA). To this end, upper bounds against which the performances can be evaluated are proposed. The experimental results on five datasets clearly show the effectiveness of the selection methods based on local accuracy estimates.
Performances of face recognition systems based on principal component analysis can degrade quickl... more Performances of face recognition systems based on principal component analysis can degrade quickly when input images exhibit substantial variations, due for example to changes in illumination or pose, compared to the templates collected during the enrolment stage. On the other hand, a lot of new unlabelled face images, which could be potentially used to update the templates and re-train the system, are made available during the system operation. In this paper a semi-supervised version, based on the self-training method, of the classical PCA-based face recognition algorithm is proposed to exploit unlabelled data for off-line updating of the eigenspace and the templates. Reported results show that the exploitation of unlabelled data by self-training can substantially improve the performances achieved with a small set of labelled training examples.
Uploads
Papers by Gian Luca Marcialis