We examine the performance of multimodal biometric authentication systems using state-of-the-art Commercial Off-the-Shelf (COTS) fingerprint and face biometric systems on a population approaching 1,000 individuals. The majority of prior studies of multimodal biometrics have been limited to relatively low accuracy non-COTS systems and populations of a few hundred users. Our work is the first to demonstrate that multimodal fingerprint and face biometric systems can achieve significant accuracy gains over either biometric alone, even when using highly accurate COTS systems on a relatively large-scale population. In addition to examining well-known multimodal methods, we introduce new methods of normalization and fusion that further improve the accuracy.

A novel kernel-based fusion strategy is presented. It is based on SVM classifiers, trade-o# coe#cients introduced in the standard SVM training and testing procedures, and quality measures of the input biometric signals. Experimental results on a prototype application based on voice and fingerprint traits are reported. The benefits of using the two modalities as compared to only using one of them are revealed. This is achieved by using a novel experimental procedure in which multi-modal verification performance tests are compared with multi-probe tests of the individual subsystems. Appropriate selection of the parameters of the proposed quality-based scheme leads to a quality-based fusion scheme outperforming the raw fusion strategy without considering quality signals. In particular, a relative improvement of 18% is obtained for small SVM training set size by using only fingerprint quality labels.

A novel adapted strategy for combining general and user-dependent knowledge at the decision-level in multimodal biometric authentication is presented. Userindependent, user-dependent, and adapted fusion and decision schemes are compared by using a bimodal system based on fingerprint and written signature. The adapted approach is shown to outperform the other strategies considered in this paper. Exploiting available information for training the fusion function is also shown to be better than using existing information for post-fusion trained decisions.

Multibiometric systems, which consolidate information from multiple biometric sources, are gaining popularity because they are able to overcome limitations such as non-universality, noisy sensor data, large intra-user variations and susceptibility to spoof attacks that are commonly encountered in unibiometric systems. In this thesis, we address two critical issues in the design of a multibiometric system, namely, fusion methodology and template security. First, we propose a fusion methodology based on the Neyman-Pearson theorem for combination of match scores provided by multiple biometric matchers. The likelihood ratio (LR) test used in the Neyman-Pearson theorem directly maximizes the genuine accept rate (GAR) at any desired false accept rate (FAR). The densities of genuine and impostor match scores needed for the LR test are estimated using finite Gaussian mixture models. We also extend the likelihood ratio based fusion scheme to incorporate the quality of the biometric samples. Further, we also show that the LR framework can be used for designing sequential multibiometric systems by constructing a binary decision tree classifier based on the marginal likelihood ratios of the

Abstract. Combining multiple biometrics may enhance the performance of personal authentication system in accuracy and reliability. In this paper, we compare 13 combination methods in the context of combining the voiceprint and fingerprint recognition system in two different modes: verification and identification. The experimental results show that Support Vector Machine and the Dempster-Shafer method are superior to other schemes. 1.

Broad classes of statistical classification algorithms have been developed and applied successfully to a wide range of real world domains. In general, ensuring that the particular classification algorithm matches the properties of the data is crucial in providing results that meet the needs of the particular application domain. One way in which the impact of this algorithm/application match can be alleviated is by using ensembles of classifiers, where a variety of classifiers (either different types of classifiers or different instantiations of the same classifier) are pooled before a final classification decision is made. Intuitively, classifier ensembles allow the different needs of a difficult problem to be handled by classifiers suited to those particular needs. Mathematically, classifier ensembles provide an extra degree of freedom in the classical bias/variance tradeoff, allowing solutions that would be difficult (if not impossible) to reach with only a single classifier. Because of these advantages, classifier ensembles have been applied to many difficult real world problems. In this paper, we survey select applications of ensemble methods to problems that have historically been most representative of the difficulties in classification. In particular, we survey applications of ensemble methods to remote sensing, person recognition, one vs. all recognition, and medicine.

Although the problem of automatic fingerprint matching has been extensively studied, it is nevertheless, not a fully solved problem. In this thesis, an information fusion approach is adopted to address some of the limitations of existing fingerprint matching systems. A hybrid fingerprint system that utilizes both minutiae points and ridge feature maps to represent and match fingerprint images has been developed. The hybrid matcher is shown to perform significantly better than a traditional minutiae-based matcher. The ridge feature maps extracted by this technique have also been used to align and register fingerprint image pairs via a correlation process, thereby obviating the need to rely on minutiae points for image registration. To address the problem of partial prints obtained from small-sized sensors, a fingerprint mosaicking scheme has been developed. The proposed technique constructs a composite fingerprint template from two partial fingerprint impressions by using the iterative control point (ICP) algorithm that determines the transformation parameters relating the two impressions. To mitigate the effect of non-linear distortionsin fingerprint images on the matching process, an average deformation model has

Automatically verifying the identity of a person by means of biometrics (e.g., face and fingerprint) is an important application in our day-to-day activities such as accessing banking services and security control in airports. To increase the system reliability, several biometric devices are often used. Such a combined system is known as a multimodal biometric system. This paper reports a benchmarking study carried out within the framework the Biosecure DS2 (Access Control) evaluation campaign organized by the University of Surrey, involving face, fingerprint and iris biometrics for person authentication, targeting the application of physical access control in a mediumsize establishment with some 500 persons. While multimodal biometrics is a well investigated subject in the literature, there exists no benchmark for a fusion algorithm comparison. Working towards this goal, we designed two sets of experiments: quality-dependent

Humans are excellent experts in person recognition and yet they do not perform excessively well in recognizing others only based on one modality such as single facial image. Experimental evidence of this fact is reported concluding that even human authentication relies on multimodal signal analysis. The elements of automatic multimodal authentication along with system models are then presented. These include the machine experts as well as machine supervisors. In particular, fingerprint and speech based systems will serve as illustration. A signal adaptive supervisor based on the input biometric signal quality is evaluated.

Fusing the scores of several biometric systems is a very promising approach to improve the overall system’s accuracy. Despite many works in the literature, it is surprising that there is no coordinated effort in making a benchmark database available. It should be noted that fusion in this context consists not only of multimodal fusion, but also intramodal fusion, i.e., fusing systems using the same biometric modality but different features, or same features but using different classifiers. Building baseline systems from scratch often prevents researchers from putting more efforts in understanding the fusion problem. This paper describes a database of scores taken from experiments carried out on the XM2VTS face and speaker verification database. It then proposes several fusion protocols and provides some state-of-the-art tools to evaluate the fusion performance.