Figure 3 - uploaded by Christophe Rosenberger
Content may be subject to copyright.
Source publication
We address the selection of fingerprint minutiae given a fingerprint ISO template. Minutiae selection plays a very important role when a secure element (i.e. a smart-card) is used. Because of the limited capability of computation and memory, the number of minutiae of a stored reference in the secure element is limited. We propose in this paper a co...
Context in source publication
Context 1
... address the selection of fingerprint minutiae given a fingerprint ISO template. Minutiae selection plays a very important role when a secure element (i.e. a smart-card) is used. Because of the limited capability of computation and memory, the number of minutiae of a stored reference in the secure element is limited. We propose in this paper a comparative study of 7 minutiae selection methods from the literature. Experimental results on 3 fingerprint databases from the Fingerprint Verification Competition show their relative efficiency in terms of performance and computation time. Keywords: Minutiae Selection, ISO fingerprint template Nowadays, electronic transactions are part of our daily life (e-commerce, smartphones, physical access control . . . ). In order to guarantee the security of authentication, biometrics is often used. Many real applications benefit from this technology such as for user access control or e-payment. Nevertheless, a biometric data is very sensitive and cannot be revoked in general (like a password). In order to ensure its security and privacy, a biometric data is usually stored in a Secure Element (SE). The Secure Element could be a SmartCard, with an (OCC) On- Card-Comparison algorithm inside. Two steps are necessary when using a biometric system, 1) the enrolment and 2) the verification as described in Figure 1. The OCC algorithm permits to compute a comparison score between a captured biometric template with the reference one. It is a common, the biometric template stored in the SE follows the ISO Compact Card standard ? to ensure the interoperability between biometric systems. This template is composed of a set of minutiae represented by 3 octets and 4 values ( x i , y i , T i , θ i ) , i = 1 : N j where ( x i , y i ) correspond to the location of the minutiae in the image, T i corresponds to the minutiae type (bifurcation, ridge ending . . . ), θ i to the minutiae orientation (related to the ridge) and N j the number of minutiae for the sample j of the user (see Figure 2). A SE has hardware and software constraints as for example the size of memory, the number of data we can send with an APDU command ? (ISO 7816 standard for the communication with a SE). These limitations have an impact on the embedded algorithm and the size of the fingerprint template. Generally, in an operational application, a fingerprint template is limited to 48 minutiae when stored in a SE. When the sensor extracts more minutiae than the OCC is able to process, we have to reduce the size of the template by selecting the most appropriate minutiae. On the state-of-the-art only few methods has been proposed. One by the ISO consortium, ? this method is based on the peeling off minutiae and we name it ISOTruncation in this paper. The second is presented by the NIST, ? this method is based on the CORE distance with all minutiae we keep only minutiae nearest the CORE point, we name it NISTBarycenter. It is tedious to determine if a method is better than another one. For this study, we are placed in the worst case on real field, we don’t have the image. We have no information about the extractor algorithm and the On-Card-Comparison algorithm. We just now the format of the template (ISO Compact Card). In our setting, we have on input the template and we send it to the OCC, if the size is too high, the OCC send us a response with the max length authorize for the comparison. To reduce the template, we used the Minutiae selection block in Figure 3, on the output we obtain a reduce template ready to send to the OCC with the good size. To summarize, the aim of this paper is to determine how the way to select the minutiae modifies the performance of the biometric authentication without knowledge on extractor and On-Card-Comparison algorithm. For this purpose, we study several selection methods present on the Minutiae selection block. In this section, we present the studied selection methods. Some come from the state of the art and we also proposed many of them. in total, we consider in this study 7 selection methods. The methods are placed in three gender, original template, reordering minutiae by truncation and minutiae clustering. The first method is to consider all minutiae in the template. The performance associated to the initial template is used as reference for the experimental results. We could expect as for example a decrease of the performance of other selection methods face to this method. This method is based on a pruning mechanism. This approach is simple and fast (few milliseconds). It has been shown that minutiae located near the core of the fingerprint are the most useful for the matching process. Given a fingerprint template, the core location is in general unknown but the centroid of minutiae is a good estimate. This minutiae selection approach tends to keep minutiae near the centroid for this reason. We have four steps in this method ...
Similar publications
Biometric recognition is already a big player in how we interact with our phones and access control systems. This is a result of its comfort of use, speed and security. For the case of border control, it eases the task of person identification and black-list checking. Although the performance rates for verification and identification have dropped i...
Citations
... Testing all combinations (selection of a subset of minutiae) is not possible for computation limitations. Few algorithms for minutiae selection have been proposed in the literature (ISO, 2007;Vibert et al., 2015Vibert et al., , 2018, and the scope of the proposed study is to compare the main methods. ...
Biometric systems are more and more used for many applications (physical access control, e-payment, etc.). Digital fingerprint is an interesting biometric modality as it can easily be used for embedded systems (smartcard, smartphone, and smartwatch). A fingerprint template is composed of a set of minutiae used for their comparison. In embedded systems, a secure element is in general used to store and compare fingerprint templates to meet security and privacy requirements. Nevertheless, it is necessary to select a subset of minutiae from a template due to storage and computation constraints. In this study, we present, a comparative study of the main minutiae selection methods from the literature. The considered methods require no further information like the raw image. Experimental results show their relative performance when using different matching algorithms and datasets. We identified that some methods can be used within different contexts (enrollment or verification) with minimal degradation of performance.
... Minutiae selection plays a vital role in minutiae based fingerprint detection techniques. In [5], seven minutiae selection methods are stated and shown that how the selection of minutiae points affects the performance of on-card-comparison algorithm. At first the ideal case, i.e. all minutiae points are considered and then six different test cases of minutiae selection under truncation based methods are represented. ...
Traditional fingerprint matching algorithms primarily focus on minutiae points on fingertip surface. In this paper, a novel approach is proposed for fingerprint matching that is based on ridge and valley characteristics of fingerprints. At first, the input fingerprint image is normalized and the registration point of that particular fingerprint is detected. Then a line profile is generated centering on that reference point. The distances between the reference point and ridges and the count of intersection points of line profile and ridges are stored in database. This process is repeated after every 15-degree angle to 345-degree in clock-wise direction and for orientation angle, the distances are stored sequentially. For matching intersection point count number along with the sequence of distance values are compared with the stored values. This new method can detect fingerprint from any orientation angle. Experimental result shows 90.87% accuracy of the proposed method.
... The ES measures a quality metric via a statistically computed global EER value, indicating the contribution of the quality metric in the degradation of the overall error rate. This section gives a description of an auxiliary means of the employed evaluation approach, which relies on the no-image minutiae selection (NIMS) [19]. The NIMS is an operation to reduce the size of minutiae template in terms of a desired number of minutiae points [19]. ...
... This section gives a description of an auxiliary means of the employed evaluation approach, which relies on the no-image minutiae selection (NIMS) [19]. The NIMS is an operation to reduce the size of minutiae template in terms of a desired number of minutiae points [19]. In this study, we use a selection criterion namely Vertex (abbr. ...
... A little bit variation appears as the desired number increased to 48 or so when calculating the global EER with NBIS matching scores. This is reasonable according to the study of NIMS [19]. In this case, with the reference quality metric and the objective measure (utility), one can found that the proposed framework is a valid solution for assessing fingerprint quality. ...
... The quality-based selection is followed by distance-based approach with respect to the image center if the remaining points still exceed the maximum minutiae number [107]. Vibert et al. [108] proposed several NIMS approaches to perform blind selection with non-compact templates. The kmeans and truncation proposed in [108] are used as the reference. ...
... Vibert et al. [108] proposed several NIMS approaches to perform blind selection with non-compact templates. The kmeans and truncation proposed in [108] are used as the reference. The former is implemented with the Fuzzy c-means [109] algorithm clustering the minutiae of one template into several groups and the points are pruned in terms of their membership grade with regard to the associated cluster(s). ...
... Therefore, we simply choose one original database as an example, for which the 04DB1 is used since there is a dissent between two matchers. In addition, the reduced datasets and their matching scores of 04DB1 are already available in [108]. The matching performance with the original template (NoSel) of this database is far from the utility-based EER (VertUtility), which makes a clear illustration, see plots of the global EERs given in Figure 3. 13. ...
La qualité d'image est un facteur important de systèmes d'identification automatique par empreintes digitales (AFIS) parce que la performance d'appariement pourrait être affectée de manière significative par des échantillons de mauvaise qualité. La qualité des échantillons d'empreintes digitales, par exemple, pourrait être grandement affectée par un grand nombre de problèmes, tels que les défauts physiques et la performance de l'appareil de détection (facteurs externes). En outre, la différence entre les données capturées telles que la spécification de l'image permet également certaines difficultés pour obtenir une solution commune. Les études existantes ont fait des efforts pour connaître l'approche la plus appropriée pour représenter la qualité de l'empreinte digitale. Cette thèse propose plusieurs méthodes d'évaluation de la qualité, tendant à obtenir des observations de cette question par plusieurs aspects : La mesure de qualité permet-elle de prédire la performance de l'appariement ? Dans quelle mesure la fusion multi-attributs permet de réaliser une bonne évaluation de la qualité d'empreintes digitales ? Est-il possible de se qualifier les empreintes digitales avec seulement son modèle de minuties ? Est-il possible d'évaluer la qualité de l'empreinte digitale via la fusion de plusieurs éléments dans la phase de segmentation ?Les approches proposées dans cette thèse sont des réponses à ces questions.
... The MINEX II [5] presents two cases for pruning minutiae points according to the image information: quality-based and distance-based approach with respect to the image center. Vibert et al. [13] proposed several NIMS approaches to perform blind selection with noncompact templates. The kmeans and truncation proposed in [13] are used as the reference. ...
... Vibert et al. [13] proposed several NIMS approaches to perform blind selection with noncompact templates. The kmeans and truncation proposed in [13] are used as the reference. The former is implemented with the Fuzzy c-means [9] algorithm clustering the minutiae of one template into several groups and the points are pruned in terms of their membership grade with regard to the associated cluster(s). ...
The embedded applications of fingerprint proposed so far are chiefly based on the minutiae template. This kind
of system is not resource-free and minutiae template is generally sacrificed to cover the shortage. This paper
presents several simple yet efficient no-image minutiae selection approaches (NIMS) for the standard minutiae templates (ISO/IEC 19794-2). With the reduced-templates obtained by using the proposed methods, the overall performance can be guaranteed in comparing with the results generated by the original templates. The interoperability tests are performed with several FVC databases. An additional analysis with the quality of the enrollment samples is also carried out. The experimental results demonstrate the validity and efficiency of the proposed approaches.
