Matching Score Level Research Articles

Multi-directional two-dimensional PCA with matching score level fusion for face recognition

The traditional matrix-based feature extraction methods that have been widely used in face recognition essentially work on the facial image matrixes only in one or two directions. For example, 2DPCA can be seen as the row-based PCA and only reflects the information in each row, and some structure information cannot be uncovered by it. In this paper, we propose the directional 2DPCA that can extract features from the matrixes in any direction. To effectively use all the features extracted by the D2DPCA, we combine a bank of D2DPCA performed in different directions to develop a matching score level fusion method named multi-directional 2DPCA for face recognition. The results of experiments on AR and FERET datasets show that the proposed method can obtain a higher accuracy than the previous matrix-based feature extraction methods.

Robust Way of Multimodal Biometric Score Normalization

Presently Multimodal biometric authentication systems provide better performance compared to unimodal systems. A Multimodal system is designed by combining the information from different biometrics. In addition, fusion at the matching score level produces superior results due to ease in accessing the scores generated by different matchers. This scheme requires score normalization to transform the scores into a common domain, because the matching scores output by various modalities are heterogeneous. Hence, we have analyzed the performance of different normalization techniques and fusion rules on a multimodal biometric system comprising of retina, ear, and fingerprint traits of a user. We also propose a new approach for normalizing the scores of traits to enhance the accuracy of biometric authentication process. The experimental results obtained using three different databases show that the use of modified Two Quadrics (2Q) and modified Quadric-line-Quadric (QLQ) normalization schemes result in a better recognition performance compared to the other methods.

Passive Multimodal 2-D+3-D Face Recognition Using Gabor Features and Landmark Distances

We introduce a novel multimodal framework for face recognition based on local attributes calculated from range and portrait image pairs. Gabor coefficients are computed at automatically detected landmark locations and combined with powerful anthropometric features defined in the form of geodesic and Euclidean distances between pairs of fiducial points. We make the pragmatic assumption that the 2-D and 3-D data is acquired passively (e.g., via stereo ranging) with perfect registration between the portrait data and the range data. Statistical learning approaches are evaluated independently to reduce the dimensionality of the 2-D and 3-D Gabor coefficients and the anthropometric distances. Three parallel face recognizers that result from applying the best performing statistical learning schemes are fused at the match score-level to construct a unified multimodal (2-D+3-D) face recognition system with boosted performance. Performance of the proposed algorithm is evaluated on a large public database of range and portrait image pairs and found to perform quite well.

Markov chain model for multimodal biometric rank fusion

Multimodal biometric aims at increasing reliability of biometric systems through utilizing more than one biometric in decision-making process. An effective fusion scheme is necessary for combining information from various sources. Such information can be integrated at several distinct levels, such as sensor level, feature level, match score level, rank level, and decision level. In this paper, we present a multimodal biometric system utilizing face, iris, and ear biometric features through rank level fusion method using novel Markov chain approach. We first apply fisherimage technique to face and ear image databases for recognition and Hough transform and Hamming distance techniques for iris image recognition. The main contribution is in introducing Markov chain approach for biometric rank aggregation. One of the distinctive features of this method is that it satisfies the Condorcet criterion, which is essential in any fair rank aggregation system. The experimentation shows superiority of the proposed approach to other recently introduced biometric rank aggregation methods. The developed system can be effectively used by security and intelligence services for controlling access to prohibited areas and protecting important national or public information.

Feature selection for support vector machine-based face-iris multimodal biometric system

Multimodal biometric can overcome the limitation possessed by single biometric trait and give better classification accuracy. This paper proposes face-iris multimodal biometric system based on fusion at matching score level using support vector machine (SVM). The performances of face and iris recognition can be enhanced using a proposed feature selection method to select an optimal subset of features. Besides, a simple computation speed-up method is proposed for SVM. The results show that the proposed feature selection method is able improve the classification accuracy in terms of total error rate. The support vector machine-based fusion method also gave very promising results.

User Identification Using Gait Patterns on UbiFloorII

This paper presents a system of identifying individuals by their gait patterns. We take into account various distinguishable features that can be extracted from a user’s gait and then divide them into two classes: walking pattern and stepping pattern. The conditions we assume are that our target environments are domestic areas, the number of users is smaller than 10, and all users ambulate with bare feet considering the everyday lifestyle of the Korean home. Under these conditions, we have developed a system that identifies individuals’ gait patterns using our biometric sensor, UbiFloorII. We have created UbiFloorII to collect walking samples and created software modules to extract the user’s gait pattern. To identify the users based on the gait patterns extracted from walking samples over UbiFloorII, we have deployed multilayer perceptron network, a feedforward artificial neural network model. The results show that both walking pattern and stepping pattern extracted from users’ gait over the UbiFloorII are distinguishable enough to identify the users and that fusing two classifiers at the matching score level improves the recognition accuracy. Therefore, our proposed system may provide unobtrusive and automatic user identification methods in ubiquitous computing environments, particularly in domestic areas.

Robust Algorithms for a Multimodal Biometric System Using Palmprint and Speech

In this paper, we propose a person verification scheme using a novel combination of palmprint and speech. The crucial aspect of biometric based verification lies in its use of features in verification. Thus, in this paper, we propose two novel feature extraction methods for palmprint verification. The proposed methods are based on the Gaussian mixture model followed by subspace based approaches such as ICA I and ICA II, called independent component analysis I mixture model (ICA I MM) and independent component analysis II mixture model (ICA II MM) for palmprint verification. The speech verification system uses sphericity based measurement for person verification. The proposed multimodal biometric system uses a fairly large database of 150 users. The robustness of the proposed methods is validated by introducing six different types of noises on palmprint images and Gaussian white noise with two different signal-to-noise ratios on speech samples. Then, fusion of palmprint and speech are carried out at match score level using the weighted SUM rule for both clean and noisy conditions. Finally, we compare the results of combined biometric with the results of individual biometric and also the results of the proposed methods against conventional (without mixture model) subspace methods.

Using Multiple Level Fusion for Improving Performance of Speaker Recognition

In this paper, a multiple level fusion framework to apply into the automatic speaker recognition system in order to improve its performance is presented. Based on the framework, different multiple level fusion methods, such as a strong multiple level fusion and three weak multiple level fusions, are defined in this paper. To examine the effectiveness of the proposed framework, two-feature combination scheme would be considered. After investigating the availability of strong and weak multiple level fusions for this scheme, the framework adopts a weak multiple level fusion method which combines two level fusions, ie matching-score fusion and decisionmaking fusion. In the matching-score level, a commonly used method called the score vector fusion is adopted. In the decision-making level, the kernel combination, also known as Multiple Kernel Learning is chosen. These two techniques can be embedded into many automatic speaker recognition systems. Throughout the evaluation by NIST 2001 corpus, two sets of exper...

Designing efficient fusion schemes for multimodal biometric systems using face and palmprint

In this paper, we address the problem of designing efficient fusion schemes of complementary biometric modalities such as face and palmprint, which are effectively coded using Log-Gabor transformations, resulting in high dimensional feature spaces. We propose different fusion schemes at match score level and feature level, which we compare on a database of 250 virtual people built from the face FRGC and the palmprint PolyU databases. Moreover, in order to reduce the complexity of the fusion scheme, we implement a particle swarm optimization (PSO) procedure which allows the number of features (identifying a dominant subspace of the large dimension feature space) to be significantly reduced while keeping the same level of performance. Results in both closed identification and verification rates show a significant improvement of 6% in performance when performing feature fusion in Log-Gabor space over the more common optimized match score level fusion method.

An Adaptive Multimodal Biometric Recognition Algorithm for Face Image using Speech Signal

A multimodal biometric authentication system based on plastic surgery face image using text dependent speech signal is described in this paper. In addition, the system is designed to keep the rate as high as possible for the plastic surgery face image by using text dependent speech signal. Each module of the system, i.e. the face and speech, is developed separately and fusion is done at matching level to obtain the optimal score for the multimodal biometric recognition system. Although information fusion in a multimodal system can be performed at various levels, integration at the matching score level is the most common approach due to the ease in accessing and combining the scores generated by different matchers. Since the matching scores output by the various modalities are heterogeneous, score normalization is needed to transform these scores into a common domain, prior to combining them.

An Efficient Process of Human Recognition Fusing Palmprint and Speech features

This paper presents fusion of two biometric traits, i.e., palmprint and speech signal, at matching score level architecture uses weighted sum of score technique. The features are extracted from the pre-processed palm image and pre-processed speech signal. The features of a query image and speech signal are compared with those of a database images and speech signal to obtain matching scores. The individual scores generated after matching are passed to the fusion module. This module consists of three major steps i.e., normalization, generation of similarity score and fusion of weighted scores. The final score is then used to declare the person as genuine or an impostor. The system is tested on database collected by the authors for 120 subjects and gives an overall accuracy of 98.47% with FAR of 1.36% and FRR of 0.87%.

Addressing Missing Values in Kernel-Based Multimodal Biometric Fusion Using Neutral Point Substitution

In multimodal biometric information fusion, it is common to encounter missing modalities in which matching cannot be performed. As a result, at the match score level, this implies that scores will be missing. We address the multimodal fusion problem involving missing modalities (scores) using support vector machines (SVMs) with the neutral point substitution (NPS) method. The approach starts by processing each modality using a kernel. When a modality is missing, at the kernel level, the missing modality is substituted by one that is unbiased with regards to the classification, called a neutral point. Critically, unlike conventional missing-data substitution methods, explicit calculation of neutral points may be omitted by virtue of their implicit incorporation within the SVM training framework. Experiments based on the publicly available Biosecure DS2 multimodal (scores) data set show that the SVM-NPS approach achieves very good generalization performance compared to the sum rule fusion, especially with severe missing modalities.

Represent and fuse bimodal biometric images at the feature level: complex-matrix-based fusion scheme

Multibiometrics can obtain a higher accuracy than the single biometrics by simultaneously using multiple biometric traits of the sub- ject. We note that biometric traits are usually in the form of images. Thus, how to properly fuse the information of multiple biometric images of the subject for authentication is crucial for multibiometrics. We propose a novel image-based linear discriminant analysis IBLDA approach to fuse two biometric traits i.e., bimodal biometric images of the same subject in the form of matrix at the feature level. IBLDA first integrates two bio- metric traits of one subject into a complex matrix and then directly ex- tracts low-dimensional features for the integrated biometric traits. IBLDA also enables more information to be exploited than the matching score level fusion and the decision level fusion. Compared to linear discrimi- nant analysis LDA, IBLDA has the following advantages: First, it can overcome the small sample size problem that conventional LDA usually suffers from. Second, IBLDA solves the eigenequation at a low compu- tational cost. Third, when storing the scatter matrices IBLDA will not bring as heavy a memory burden as conventional LDA. We also clearly show the theoretical foundation of the proposed method. The experiment result shows that the proposed method can obtain a high classification accuracy. © 2010 Society of Photo-Optical Instrumentation Engineers.

An Enhanced Palm Vein Recognition System Using Multi-level Fusion of Multimodal Features and Adaptive Resonance Theory

An improved palm vein recognition system using multimodal features and neural network classifier has been developed and presented in this paper. The effects of fusion of multiple features at various levels have been demonstrated. The shape and texture features have been considered for recognition of authenticated users and it is validated using neural network classifier. The recognition accuracy of the proposed system has been compared with the existing techniques. It is found that the recognition accuracy is 99.61% when the multimodal features fused at matching score level. This proposed multimodal palm vein recognition system is expected to provide reliable security.

Integration of multiple cues for a human authentication system

In the field of biometrics, personal authentication and identification are regarded as effective methods for automatic recognition of a person. Using multimodal biometric systems, we typically get better recognition performance compared to using a single biometric modality. We propose a multimodal biometric system using two modalities: palmprint and speech. Integrating the palmprint and speech features increases the recognition performance. We extract the discriminant features using a modified canonical form method for the Palmprint and the Mel Frequency Cepstral Coefficients (MFCC) technique for speech. The final decision is then made by fusion at the matching score level. Using a large database as the test set, the experimental results show significant improvement in reliability of recognition systems and demonstrate increases in the recognition rate.

An intelligent multimodal biometric system for high security access

The interface of computer technologies and biology is having a huge impact on society. Human recognition technology research projects promises new life to many security consulting firms and personal identification system manufacturers. The physiological or biological features of an individual, called biometrics, are unique to each human and remain unaltered during a person's lifetime and give a promising solution for security management. This paper presents an intelligent multimodal biometric verification system for physical access control, based on fusion of iris, face and fingerprint patterns. Feature vectors are created independently for query images and are then compared with the enrolled templates of each biometric trait to compute the matching score. The final decision is made by fusion at this matching score level. The proposed system is designed to suit embedded solutions for high security access in pervasive environments using biometric features.

A novel technique for palmprint classification and authentication

Palmprint database is categorised into two groups, right hand group and left hand group and then each group is classified based on principal line i.e., Heart Line. A rectangular Region of Interest (ROI), in which only heart line is present, is extracted for classification, which is divided into six regions and the palmprint database is classified based on the regions that the heart line traverses in the palm. ROI is extracted for the authentication, using this, techniques like filiformity to extract the line like features and Gabor filter to extract texture features from the palm are used. Performance of the two techniques determined individually. A sum rule is applied to combine the features obtained from the two techniques to develop an intramodel system. Fusion is applied at both feature level as well as matching score level for the authentication mechanism.

Combination of face and gait for human recognition

Focusing on the application of intelligent surveillance,this paper proposed a new approach in which the combination of face and gait was used for human recognition at a distance in video sequences. Hidden Markov Model (HMM) and Fisher faces method were primarily applied for gait and face recognition,respectively. And then,the results obtained from the two classifiers were utilized and integrated at match score level. The system was tested on video sequences of 31 individuals collected from different directions. The results show that the combination of face and gait provides a more robust recognition strategy,and it has better recognition performance compared with face-only or gait-only method.

A feature extraction method for use with bimodal biometrics

Bimodal biometrics has been found to outperform single biometrics and are usually implemented using the matching score level or decision level fusion, though this fusion will enable less information of bimodal biometric traits to be exploited for personal authentication than fusion at the feature level. This paper proposes matrix-based complex PCA (MCPCA), a feature level fusion method for bimodal biometrics that uses a complex matrix to denote two biometric traits from one subject. The method respectively takes the two images from two biometric traits of a subject as the real part and imaginary part of a complex matrix. MCPCA applies a novel and mathematically tractable algorithm for extracting features directly from complex matrices. We also show that MCPCA has a sound theoretical foundation and the previous matrix-based PCA technique, two-dimensional PCA (2DPCA), is only one special form of the proposed method. On the other hand, the features extracted by the developed method may have a large number of data items (each real number in the obtained features is called one data item). In order to obtain features with a small number of data items, we have devised a two-step feature extraction scheme. Our experiments show that the proposed two-step feature extraction scheme can achieve a higher classification accuracy than the 2DPCA and PCA techniques.

Information fusion in personal biometric authentication based on the iris pattern

Information fusion in biometrics has received considerable attention. This paper focuses on the application of information fusion techniques in iris recognition. To improve the reliability and accuracy of personal identification based on the iris pattern, this paper proposes the schemes of multialgorithmic fusion and multiinstance fusion. Multialgorithmic fusion integrates the improved phase algorithm and the DCT-based algorithm, and multiinstance fusion combines information from the left iris and the right iris of an individual. Both multialgorithmic fusion and multiinstance fusion are carried out at the matching score level and the support vector machine (SVM)-based fusion rule is utilized to generate fused scores for final decision. The experimental results on the noisy iris database UBIRIS demonstrate that the proposed fusion schemes can perform better than the single recognition systems, and further prove that information fusion techniques are feasible and effective to improve the accuracy and robustness of iris recognition especially under noisy conditions.