Search

Active Shape Model (ASM) has proved to be a powerful tool for interpreting face images. However, it may fail in the presence of non-Gaussian noise, or outliers. In this paper, we present a framework for both automatic model construction and efficient model fitting with outliers. In model construction, the training face samples are automatically labeled by local image search using Gabor wavelet features. Then robust principle component analysis (RPCA) is applied to capture the statistics of shape variations. In model fitting, an error function is introduced to deal with the outlier problem, which provides a connection to robust M-estimation. Gauss-Newton algorithm is adopted to efficiently optimize the robust energy function. Extensive experiments demonstrate the efficiency and robustness of our approach over previous methods. [ABSTRACT FROM AUTHOR]

Human gait properties can be affected by various environmental contexts such as walking surface and carrying objects. In this paper, we propose a novel approach for individual recognition by combining different gait classifiers with the knowledge of environmental contexts to improve the recognition performance. Different classifiers are designed to handle different environmental contexts, and context specific features are explored for context characterization. In the recognition procedure, we can determine the probability of environmental contexts in any probe sequence according to its context features, and apply the probabilistic classifier combination strategies for the recognition. Experimental results demonstrate the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

Content-based face image retrieval is concerned with computer retrieval of face images (of a given subject) based on the geometric or statistical features automatically derived from these images. It is well known that color spaces provide powerful information for image indexing and retrieval by means of color invariants, color histogram, color texture, etc.. This paper assesses comparatively the performance of content-based face image retrieval in different color spaces using a standard algorithm, the Principal Component Analysis (PCA), which has become a popular algorithm in the face recognition community. In particular, we comparatively assess 12 color spaces (RGB, HSV, YUV, YCbCr, XYZ, YIQ, L*a*b*, U*V*W*, L*u*v*, I1I2I3, HSI, and rgb) by evaluating 7 color configurations for every single color space. A color configuration is defined by an individual or a combination of color component images. Take the RGB color space as an example, possible color configurations are R, G, B, RG, RB, GB, and RGB. Experimental results using 1,800 FERET R, G, B images corresponding to 200 subjects show that some color configurations, such as R in the RGB color space and V in the HSV color space, help improve face retrieval performance. [ABSTRACT FROM AUTHOR]

This paper presents a 3D face recognition system based on geometrically localized facial features. We propose the feature extraction procedure using the geometrical characteristics of a face. We extract three curvatures, eight invariant facial feature points and their relative features. These features are directly applied to face recognition algorithms which are a depth-based DP (Dynamic Programming) and a feature-based SVM (Support Vector Machine). Experimental results show that face recognition rates based on the depth-based DP and the feature-based SVM are 95% for 20 people and 96% for 100 people, respectively. [ABSTRACT FROM AUTHOR]

In this paper, an investigation is carried out regarding combination of 3D and 2D information for face recognition. A two-stage method, PCA and a Reduced Multivariate Polynomial Model (RMPM), is developed to fuse the appearance and depth information of face images in feature level where simplicity (number of polynomial coefficients increases linearly with model-order and input-dimension, i.e. no dimension explosion as in the case of full multivariate polynomials)and ease of use (can be easily formulated into recursive learning fashion) are major concerns. To cater for fast on-line registration capability when a new user arrives, the learning is formulated into recursive form. The improvement of the face recognition rate using this combination is quantified. The recognition rate by the combination is better than either appearance alone or depth alone. The performance of the algorithm is verified on both XM2VTS database and a real-time stereo vision system, showing that it is able to detect, track and recognize a person walking towards a stereo camera within reasonable time. [ABSTRACT FROM AUTHOR]

The issues of fusion with client-dependent and confidence information have been well studied separately in biometric authentication. In this study, we propose to take advantage of both sources of information in a discriminative framework. Initially, each source of information is processed on a per expert basis (plus on a per client basis for the first information and on a per example basis for the second information). Then, both sources of information are combined using a second-level classifier, across different experts. Although the formulation of such two-step solution is not new, the novelty lies in the way the sources of prior knowledge are incorporated prior to fusion using the second-level classifier. Because these two sources of information are of very different nature, one often needs to devise special algorithms to combine both information sources. Our framework that we call "Prior Knowledge Incorporation" has the advantage of using the standard machine learning algorithms. Based on 10 × 32=320 intramodal and multimodal fusion experiments carried out on the publicly available XM2VTS score-level fusion benchmark database, it is found that the generalisation performance of combining both information sources improves over using either or none of them, thus achieving a new state-of-the-art performance on this database. [ABSTRACT FROM AUTHOR]

Ensuring the security of medical records is becoming an increasingly important problem as modern technology is integrated into existing medical services. As a consequence of the adoption of electronic medical records in the health care sector, it is becoming more and more common for a health professional to edit and view a patient's record using a tablet PC. In order to protect the patient's privacy, as required by governmental regulations in the United States, a secure authentication system to access patient records must be used. Biometric-based access is capable of providing the necessary security. On-line signature and voice modalities seem to be the most convenient for the users in such authentication systems because a tablet PC comes equipped with the associated sensors/hardware. This paper analyzes the performance of combining the use of on-line signature and voice biometrics in order to perform robust user authentication. Signatures are verified using the dynamic programming technique of string matching. Voice is verified using a commercial, off the shelf, software development kit. In order to improve the authentication performance, we combine information from both on-line signature and voice biometrics. After suitable normalization of scores, fusion is performed at the matching score level. A prototype bimodal authentication system for accessing medical records has been designed and evaluated on a small truly multimodal database of 50 users, resulting in an average equal error rate (EER) of 0.86%. [ABSTRACT FROM AUTHOR]

The goal of a biometric encryption system is to embed a secret into a biometric template in a way that can only be decrypted with a biometric image from the enroled person. This paper describes a potential vulnerability in such systems that allows a less-than-brute force regeneration of the secret and an estimate of the enrolled image. This vulnerability requires the biometric comparison to "leak" some information from which an analogue for a match score may be calculated. Using this match score value, a "hill-climbing" attack is performed against the algorithm to calculate an estimate of the enrolled image, which is then used to decrypt the code. Results are shown against a simplified implementation of the algorithm of Soutar et al. (1998). [ABSTRACT FROM AUTHOR]

The traditional approach to evaluating the performance of a behavioral biometric such as handwriting or speech is to conduct a study involving human subjects (naïve and/or skilled "forgers") and report the system's False Reject Rate (FRR) and False Accept Rate (FAR). In this paper, we examine a different and perhaps more ominous threat: the possibility that the attacker has access to a generative model for the behavior in question, along with information gleaned about the targeted user, and can employ this in a methodical search of the space of possible inputs to the system in an attempt to break the biometric. We present preliminary experimental results examining the effectiveness of this line of attack against a published technique for constructing a biometric hash based on online handwriting data. Using a concatenative approach followed by a feature space search, our attack succeeded 49% of the time. [ABSTRACT FROM AUTHOR]

In this paper, the use of biometric systems in forensic applications is reviewed. Main differences between the aim of commercial biometric systems and forensic reporting are highlighted, showing that commercial biometric systems are not suited to directly report results to a court of law. We propose the use of a Bayesian approach for forensic reporting, in which the forensic scientist has to assess a meaningful value, in the form of a likelihood ratio (LR). This value assist the court in their decision making in a clear way, and can be computed using scores coming from any biometric system, with independence of the biometric discipline. LR computation in biometric systems is reviewed, and statistical assumptions regarding estimations involved in the process are addressed. The paper is focused in handling small sample size effects in such estimations, presenting novel experiments using a fingerprint and a voice biometric system. [ABSTRACT FROM AUTHOR]

In this paper, we investigate fusion methods for multimodal identification using several unimodal identification results. One fingerprint identification system and two face identification systems are used as fusion sources. We discuss rank level and score level fusion methods. Whereas the latter combines similarity scores, the other one combines the orders of the magnitudes of the similarity scores. For rank level methods, Borda Count and Bayes Fuse are considered and, for score level methods, Sum Rule and Binary Classification Approach are considered. Especially, we take a more detailed look at Binary Classification Approach, which simplifies a multiple class problem into a binary class problem. Finally, we compare experimental results using the fusion methods in different combinations of the sources. [ABSTRACT FROM AUTHOR]

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. [ABSTRACT FROM AUTHOR]

A multimodal biometric system integrates information from multiple biometric sources to compensate for the limitations in performance of each individual biometric system. We propose an optimal framework for combining the matching scores from multiple modalities using the likelihood ratio statistic computed using the generalized densities estimated from the genuine and impostor matching scores. The motivation for using generalized densities is that some parts of the score distributions can be discrete in nature; thus, estimating the distribution using continuous densities may be inappropriate. We present two approaches for combining evidence based on generalized densities: (i) the product rule, which assumes independence between the individual modalities, and (ii) copula models, which consider the dependence between the matching scores of multiple modalities. Experiments on the MSU and NIST multimodal databases show that both fusion rules achieve consistently high performance without adjusting for optimal weights for fusion and score normalization on a case-by-case basis. Keywords: Biometric recognition, multimodal biometric systems, fusion, Gaussian copula models, Generalized densities, Neyman-Pearson theorem. [ABSTRACT FROM AUTHOR]

In 3D face recognition systems, 3D facial shape information plays an important role. Various shape representations have been proposed in the literature. The most popular techniques are based on point clouds, surface normals, facial profiles, and statistical analysis of depth images. The contribution of the presented work can be divided into two parts: In the first part, we have developed face classifiers which use these popular techniques. A comprehensive comparison of these representation methods are given using 3D RMA dataset. Experimental results show that the linear discriminant analysis-based representation of depth images and point cloud representation perform best. In the second part of the paper, two different multiple-classifier architectures are developed to fuse individual shape-based face recognizers in parallel and hierarchical fashions at the decision level. It is shown that a significant performance improvement is possible when using rank-based decision fusion in ensemble methods. [ABSTRACT FROM AUTHOR]

In this paper, we present the results of our investigation of the use of the visual characteristics of human hair as a primary recognition attribute for human ID in indoor video imagery. The emerging need for unobtrusive biometrics has led to recent research interest in using the features of the face, gait, voice, and clothes, among others, for human authentication. However, the characteristics of hair have been almost completely excluded as a recognition attribute from state-of-the-art authentication methods. We contend that people often use hair as a principal visual biometric. Furthermore, hair is the part of the human body most likely to be visible to overhead surveillance cameras free of occlusion. Although hair can hardly be trusted to be a reliable long-term indicator of human identity, we show that the visual characteristics of hair can be effectively used to unobtrusively re-establish human ID in the task of short-term recognition and reacquisition in a video-based multiple-person continuous tracking application. We propose new pixel-based and line-segment-based features designed specifically to characterize hair, and recognition schemes that use just a few training images per subject. Our results demonstrate the feasibility of this approach, which we hope can form a basis for further research in this area. [ABSTRACT FROM AUTHOR]

With the recent emphasis on homeland security, there is an increased interest in accurate and non-invasive techniques for face recognition. Most of the current techniques perform a structural analysis of facial features from still images. Recently, video-based techniques have also been developed but they suffer from low image-quality. In this paper, we propose a new method for face recognition, called Digital Image Skin Correlation (DISC), which is based on dynamic instead of static facial features. DISC tracks the motion of skin pores on the face during a facial expression and obtains a vector field that characterizes the deformation of the face. Since it is almost impossible to imitate another person's facial expressions these deformation fields are bound to be unique to an individual. To test the performance of our method in face recognition scenarios, we have conducted experiments where we presented individuals wearing heavy make-up as disguise to our DISC matching framework. The results show superior face recognition performance when compared to the popular PCA+ LDA method, which is based on still images. [ABSTRACT FROM AUTHOR]

We introduce a novel technique for face recognition by using 3D face data that has been reconstructed from a stereo camera system. The face data consists of a dense 3D mesh of vertices describing the facial shape and geometry as well as a 2D texture map describing the facial appearance of each subject. We propose a recognition algorithm based on two steps: The first step involves the cleaning of the facial surface followed by a registration and normalization to a standard template face. The second step involves the creation of a PCA model and the use of a reduced dimensionality face space for the calculation of facial similarity. We use this technique on 3D surface and texture data comprising 83 subjects. Our results demonstrate the wealth of 3D information on the face as well as the importance of standardization and noise elimination in the datasets. [ABSTRACT FROM AUTHOR]

In this paper, we propose a method of authenticating corrupted photo images based on noise parameter estimation and implement an authentication system using TMS320C6711 DSP chip. The proposed method first generates corrupted images and the noise parameters in the training phase. With a corrupted image and an original image, the noise parameters of the corrupted photo image can be estimated in the testing phase. Finally, we can make a synthesized photo image from the original photo image using the estimated noise parameters. We made some experiments on the prototype of the stand-alone system to verify the performance of the proposed method and to apply for real-life applications. The experimental results on this system show that the proposed method can estimate the noise parameters accurately and improve the performance of photo image authentication. [ABSTRACT FROM AUTHOR]

A probabilistic system that retrieves face images based on verbal descriptions given by users is proposed. This interactive system prompts the user at each stage of query to provide a description about a facial feature that will help it to retrieve the required face image. It is a soft biometric system that groups people by the description of their faces. The method proposed automates the process of extracting general verbal descriptions of faces like 'long nosed' or 'blonde haired' and performs queries on them. The proposed method uses Bayesian learning for the query process and hence is more immune to errors in both the extraction of semantic descriptions and user given information. It was found that the required face image appeared 77.6% of the time within the top 5 and 90.4% of the time within the top 10 retrieved face images. [ABSTRACT FROM AUTHOR]

Pose and illumination changes from picture to picture are two main barriers toward full automatic face recognition. In this paper, a novel method to handle both pose and lighting condition simultaneously is proposed, which calibrates the pose and lighting condition to a pre-set reference condition through an illumination invariant 3D face reconstruction. First, some located facial landmarks and a priori statistical deformable 3D model are used to recover an elaborate 3D shape. Based on the recovered 3D shape, the "texture image" calibrated to a standard illumination is generated by spherical harmonics ratio image and finally the illumination independent 3D face is reconstructed completely. The proposed method combines the strength of statistical deformable model to describe the shape information and the compact representations of the illumination in spherical frequency space, and handle both the pose and illumination variation simultaneously. This algorithm can be used to synthesize virtual views of a given face image and enhance the performance of face recognition. The experimental results on CMU PIE database show that this method can significantly improve the accuracy of the existed face recognition method when pose and illumination are inconsistent between gallery and probe sets. [ABSTRACT FROM AUTHOR]

A novel nonlinear discriminant analysis method, Kernelized Decision Boundary Analysis (KDBA), is proposed in our paper, whose Decision Boundary feature vectors are the normal vector of the optimal Decision Boundary in terms of the Structure Risk Minimization principle. We also use a simple method to prove a property of Support Vector Machine (SVM) algorithm, which is combined with the optimal Decision Boundary Feature matrix to make our method consistent with the Kernel Fisher method(KFD). Moreover, KDBA is easily used in its applications, and the traditional Decision Boundary Analysis implementations are computationally expensive and sensitive to the size of the problem. Text classification problem is first used to testify the effectiveness of KDBA. Then experiments on the large-scale face database, the CAS-PEAL database, have illustrated its excellent performance compared with some popular face recognition methods such as Eigenface, Fisherface, and KFD. Keywords: Face Recognition, Kernel Fisher, Support Vector Machine [ABSTRACT FROM AUTHOR]

To make person authentication systems be more useful and practical, we have developed an identification algorithm, and also showed that the authentication accuracy depends on response performance. Current identification algorithm employs a comparison computation function that is optimized for one-to-one comparison. By optimizing a comparison computation function, however, it might be possible to improve response performance. In this paper, we describe design guidelines for a comparison computation function for improving the response performance of the identification. To show the guidelines, we clarify the relation between the characteristics of a matching score distribution and response performance using a matching score generation model, and also demonstrate the effectiveness of the design guidelines with a simulation using an example of another comparison computation function. [ABSTRACT FROM AUTHOR]

This paper presents a novel approach to face recognition, named Multi-resolution Histograms of Local Variation Patterns (MHLVP), in which face images are represented as the concatenation of the local spatial histogram of local variation patterns computed from the multi-resolution Gabor features. For a face image with abundant texture and shape information, aGabor feature map(GFM) is computed by convolving the image with each of the forty multi-scale and multi-orientation Gabor filters. Each GFM is then divided into small non-overlapping regions to enhance its shape information, and then Local Binary Pattern (LBP) histograms are extracted for each region and concatenated into a feature histogram to enhance the texture information in the specific GFM. Further more, all of the feature histograms extracted from the forty GFMs are further concatenated into a single feature histogram as the final representation of the given face image. Eventually, the identification is achieved by histogram intersection operation. Our experimental results on FERET face databases show that the proposed method performs terrifically better than the performance of some classical results including the best results in FERET'97. [ABSTRACT FROM AUTHOR]

This paper introduces a novel 2D Gabor-Fisher Classifier for face recognition. The 2D-GFC method applies the 2D Fisher Linear Discriminant Analysis (2D-LDA) to the gaborfaces which is derived from the Gabor wavelets representation of face images. In our method, Gabor wavelets first derive desirable facial features characterized by spatial frequency, spatial locality, and orientation selectivity to cope with the variations due to illumination and facial expression changes. 2D-LDA is then used to enhance the face recognition performance by maximizing the Fisher's linear projection criterion. To evaluate the performance of 2D-GFC, experiments were conducted on FERET database with several other methods. [ABSTRACT FROM AUTHOR]

This paper is framed in the field of statistical face analysis. In particular, the problem of accurate segmentation of prominent features of the face in frontal view images is addressed. Our method constitutes an extension of Cootes et al. [6] linear Active Shape Model (ASM) approach, which has already been used in this task [9]. The technique is built upon the development of a non-linear appearance model, incorporating a reduced set of differential invariant features as local image descriptors. These features are invariant to rigid transformations, and a subset of them is chosen by Sequential Feature Selection (SFS) for each landmark and resolution level. The new approach overcomes the unimodality and gaussianity assumptions of classical ASMs regarding the distribution of the intensity values across the training set. Validation of the method is presented against the linear ASM and its predecesor, the Optimal Features ASM (OF-ASM) [14] using the AR and XM2VTS databases as testbed. [ABSTRACT FROM AUTHOR]

In the current development of deploying biometric methods, face recognition takes a leading role. Therefore it is of interest what barriers might arise in using face recognition for certain applications. Because of the wide range of possible uses of biometrics and face recognition a technology evaluation on which image disturbances influence face recognition algorithms was done in the project BioFace III. This includes tests regarding head rotation, contrast, compression and more. [ABSTRACT FROM AUTHOR]

Many researches in face recognition have been dealing with the challenge of the great variability in head pose, lighting intensity and direction,facial expression, and aging. The main purpose of this overview is to describe the recent 3D face recognition algorithms. The last few years more and more 2D face recognition algorithms are improved and tested on less than perfect images. However, 3D models hold more information of the face, like surface information, that can be used for face recognition or subject discrimination. Another major advantage is that 3D face recognition is pose invariant. A disadvantage of most presented 3D face recognition methods is that they still treat the human face as a rigid object. This means that the methods aren't capable of handling facial expressions. Although 2D face recognition still seems to outperform the 3D face recognition methods, it is expected that this will change in the near future. [ABSTRACT FROM AUTHOR]

This paper presents a novel technique for improving face recognition performance by predicting system failure, and, if necessary, perturbing eye coordinate inputs and repredicting failure as a means of selecting the optimal perturbation for correct classification. This relies on a method that can accurately identify patterns that can lead to more accurate classification, without modifying the classification algorithm itself. To this end, a neural network is used to learn 'good' and 'bad' wavelet transforms of similarity score distributions from an analysis of the gallery. In production, face images with a high likelihood of having been incorrectly matched are reprocessed using perturbed eye coordinate inputs, and the best results used to "correct" the initial results. The overall approach suggest a more general approach involving the use of input perturbations for increasing classifier performance in general. Results for both commercial and research face-based biometrics are presented using both simulated and real data. The statistically significant results show the strong potential for this to improve system performance, especially with uncooperative subjects. [ABSTRACT FROM AUTHOR]

With the mass deployment of cameras, concern has risen about protecting a person's privacy as he goes about his daily life. Many of the cameras are installed to perform surveillance tasks that do not require the identity of a person. In the context of surveillance applications, we examine the trade-off between privacy and security. The trade-off is accomplished by looking at quantitative measures of privacy and surveillance performance. To provide privacy protection we examine the effect on surveillance performance of a parametric family of privacy function. A privacy function degrades images to make identification more difficult. By varying the parameter, different levels of privacy protection are provided. We introduce the privacy operating characteristic (POC) to quantitatively show the resulting trade-off between privacy and security. From a POC, policy makers can select the appropriate operating point for surveillance systems with regard to privacy. [ABSTRACT FROM AUTHOR]

The verification performance of biometric systems is normally evaluated using the receiver operating characteristic (ROC) or detection error trade-off (DET) curve. We propose two new ideas for statistical evaluation of biometric systems based on these data. The first is a new way to normalize match score distributions. A normalized match score, $\hat{t}$, is calculated as a function of the angle from a representation of (FMR, FNMR) values in polar coordinates from some center. This has the advantage that it does not produce counterintuitive results for systems with unusual DET performance. Secondly, building on this normalization we develop a methodology to calculate an average DET curve. Each biometric system is represented in terms of $\hat{t}$ to allow genuine and impostor distributions to be combined, and an average DET is then calulated from these new distributions. We then show that this method is equivalent to direct averaging of DET data along each angle from the center. This procedure is then applied to data from a study of human matchers of facial images. [ABSTRACT FROM AUTHOR]

Currently biometric system performance is evaluated in terms of its FAR and FRR. The accuracy expressed in such a manner depends on the characteristics of the dataset on which the system has been tested. Using different datasets for system evaluation makes a true comparison of such systems difficult, more so in cases where the systems are designed to work on different biometrics, such as fingerprint and signature. We propose a similarity metric, calculated for a pair of fingerprint templates, which will quantize the "confusion" of a fingerprint matcher in evaluating them. We show how such a metric, can be calculated for an entire biometric database, to give us the amount of difficulty a matcher has, when evaluating fingerprints from that database. This similarity metric can be calculated in linear time, making it suitable for large datasets. [ABSTRACT FROM AUTHOR]

This paper concerns the generalizability of biometric identification results from small-sized closed systems to larger open systems. Many researchers have claimed high identification accuracies on closed system consisting of a few hundred or thousand members. Here, we consider what happens to these closed identification systems as they are opened to non-members. We claim that these systems do not generalize well as the non-member population increases. To support this claim, we present experimental results on writer and iris biometric databases using Support Vector Machine (SVM) and Nearest Neighbor (NN) classifiers. We find that system security (1-FAR) decreases rapidly for closed systems when they are tested in open-system mode as the number of non members tested increases. We also find that, although systems can be trained for greater closed-system security using SVM rather than NN classifiers, the NN classifiers are better for generalizing to open systems due to their superior capability of rejecting non-members. [ABSTRACT FROM AUTHOR]

Biometric identification has emerged as a reliable means of controlling access to both physical and virtual spaces. In spite of the rapid proliferation of large-scale databases, the research has thus far been focused only on accuracy within small databases. However, as the size of the database increases, not only does the response time deteriorate, but so does the accuracy of the system. Thus for larger applications it is essential to prune the database to a smaller fraction which would not only ensure higher speeds, but also aid in achieving higher accuracy. Unlike structured information such as text or numeric data that can be sorted, biometric data does not have a natural sorting order making indexing the biometric database a challenging problem. In this paper we show the efficacy of indexing hand geometry biometric using the Pyramid Technique, to reduce the search space to just 8.86% of the entire database, while maintaining a 0% FRR. [ABSTRACT FROM AUTHOR]

The information of the vocal tract and the glottis are two kinds of sources which can characterize speakers. Though the former one has archived quite good performance in automatic speaker recognition (ASR) tasks, the glottal information behaves poorly when used individually. This work explores how to combining vocal tract and glottal information in an efficient and effective way. Taking into account the short-term correlation between them, our improved joint probability function model of the corresponding features is first proposed. Then we present a novel integrating system which uses parallel Gaussian Mixture Models (GMM) grounded on this function. Together with the traditional GMM, it also forms a hybrid model. Both methods were applied to YOHO and SRMC corpus, and experimental works show promising results. [ABSTRACT FROM AUTHOR]

Most of the prior work in biometric literature has only emphasized on the issue of feature extraction and classification. However, the critical issue of examining the usefulness of extracted biometric features has been largely ignored. Feature evaluation/selection helps to identify and remove much of the irrelevant and redundant features. The small dimension of feature set reduces the hypothesis space, which is critical for the success of online implementation in personal recognition. This paper focuses on the issue of feature subset selection and its effectiveness in a typical bimodal biometric system. The feature level fusion has not received adequate attention in the literature and therefore the performance improvement in feature level fusion using feature subset selection is also investigated. Our experimental results demonstrate that while majority of biometric features are useful in predicting the subjects identity, only a small subset of these features are necessary in practice for building an accurate model for identification. The comparison and combination of features extracted from hand images is evaluated on the diverse classification schemes; naive Bayes (normal, estimated, multinomial), decision trees (C4.5, LMT), k-NN, SVM, and FFN. [ABSTRACT FROM AUTHOR]

An audio-visual speaker identification system is described, where the audio and visual speech modalities are fused by an automatic unsupervised process that adapts to local classifier performance, by taking into account the output score based reliability estimates of both modalities. Previously reported methods do not consider that both the audio and the visual modalities can be degraded. The visual modality uses the speakers lip information. To test the robustness of the system, the audio and visual modalities are degraded to emulate various levels of train/test mismatch; employing additive white Gaussian noise for the audio and JPEG compression for the visual signals. Experiments are carried out on a large augmented data set from the XM2VTS database. The results show improved audio-visual accuracies at all tested levels of audio and visual degradation, compared to the individual audio or visual modality accuracies. For high mismatch levels, the audio, visual, and auto-adapted audio-visual accuracies are 37.1%, 48%, and 71.4% respectively. [ABSTRACT FROM AUTHOR]

The performance of deployed audio, face, and multi-modal person recognition systems in non-controlled scenarios, is typically lower than systems developed in highly controlled environments. With the aim to facilitate the development of robust audio, face, and multi-modal person recognition systems, the new large and realistic multi-modal (audio-visual) VALID database was acquired in a noisy "real world" office scenario with no control on illumination or acoustic noise. In this paper we describe the acquisition and content of the VALID database, consisting of five recording sessions of 106 subjects over a period of one month. Speaker identification experiments using visual speech features extracted from the mouth region are reported. The performance based on the uncontrolled VALID database is compared with that of the controlled XM2VTS database. The best VALID and XM2VTS based accuracies are 63.21% and 97.17% respectively. This highlights the degrading effect of an uncontrolled illumination environment and the importance of this database for deploying real world applications. The VALID database is available to the academic community through http://ee.ucd.ie/validdb/. [ABSTRACT FROM AUTHOR]

In this paper, a class of VQ-based discriminative kernel is proposed for speaker identification. Vector quantization is a well known method in speaker recognition, but its performance is not superior. The distortion of an utterance is accumulated, but the distortion source distribution on the codebook is discarded. We map an utterance to a vector by adopting the distribution and the average distortions on every code vector. Then the SVMs are used for classification. A one-versus-rest fashion is used for the problem of multiple classifications. Results on YOHO in text-independent case show that the method can improve the performance greatly and is comparative with the VQ and the basic GMM's performances. [ABSTRACT FROM AUTHOR]

DWT domain on-line signature verification method has bee proposed. Time-varying pen-position signal is decomposed into sub-band signals by using the DWT. Individual features are extracted as high frequency signals in sub-band. By using the extracted feature, verification is achieved at each sub-band and then total decision is done by combining such verification results. In this paper, we introduce a user weighting fusion into the total decision for improving verification performance. Through many verification experiments, it is confirmed that there is an optimal weight combination for each user and verifiaction rate can be improved when the optimal weight combination is applied. Such the optimal weight combination also becomes an individual feature which can not be known by others. [ABSTRACT FROM AUTHOR]

Gait as a motion-based biometric has the merit of being non-contact and unobtrusive. In this paper, we proposed a gait recognition approach using spectral features of horizontal and vertical movement of ankles in a normal walk. Gait recognition experiments using the spectral features in term of the magnitude, phase and phase-weighted magnitude show that both magnitude and phase spectra are effective gait signatures, but magnitude spectra are slightly superior. We also proposed the use of geometrical mean based spectral features for gait recognition. Experimental results with 9 subjects show encouraging results in the same-day test, while the effect of time covariate is confirmed in the cross-month test. [ABSTRACT FROM AUTHOR]

In this paper, we propose a new method for palmprint recognition based on the registration results between two palmprint images. After preprocessing, a unified coordinate system is constructed for each palmprint image. If two palmprints images are captured from the same hand, the translation, rotation and scaling alignment differences between their corresponding image blocks should be small. Therefore the registration parameters obtained by the Fourier-Mellin Transformation are used to measure the similarity between two palmprint patterns. Experimental results demonstrate the effectiveness of our ideas. [ABSTRACT FROM AUTHOR]

In this paper, we propose a driver identification method that is based on the driving behavior signals that are observed while the driver is following another vehicle. Driving behavior signals, such as the use of of the accelerator pedal, brake pedal, vehicle velocity, and distance from the vehicle in front, are measured using a driving simulator. We compared the identification rate obtained using different identification models and different features. As a result, we found the non-parametric models to be better than the parametric models. Also, the driver's operation signals were found to be better than road environment signals and car behavior signals. [ABSTRACT FROM AUTHOR]

Existing ear recognition approaches do not give theoretical or experimental performance prediction. Therefore, the discriminating power of ear biometric for human identification cannot be evaluated. This paper addresses two interrelated problems: (a) proposes an integrated local descriptor for representation to recognize human ears in 3D. Comparing local surface descriptors between a test and a model image, an initial correspondence of local surface patches is established and then filtered using simple geometric constraints. The performance of the proposed ear recognition system is evaluated on a real range image database of 52 subjects. (b) A binomial model is also presented to predict the ear recognition performance. Match and non-matched distances obtained from the database of 52 subjects are used to estimate the distributions. By modeling cumulative match characteristic (CMC) curve as a binomial distribution, the ear recognition performance can be predicted on a larger gallery. [ABSTRACT FROM AUTHOR]

This paper proposes an algorithm for generating a super-template from multiple fingerprint impressions in fingerprint enrollment for the purpose of increasing recognition accuracy. The super-template is considered as a single fingerprint template which contains highly likely true minutiae based on multiple fingerprint images. The proposed algorithm creates the super-template, in which the credibility of each minutia is updated by applying a successive Bayesian estimation (SBE) to a sequence of templates obtained from input fingerprint images. Consequently, the SBE assigns a higher credibility to frequently detected minutiae and a lower credibility to minutiae that are rarely found from the input templates. Likewise, the SBE is able to estimate the credibility of the minutia type (ridge ending or bifurcation). Preliminary experiments demonstrate that, as the number of fingerprint images increases, the proposed algorithm can improve the recognition performance, while keeping the processing time and memory storage required for the super-template almost constant. [ABSTRACT FROM AUTHOR]

Biometrics-based authentication systems offer enhanced security and user convenience compared to traditional token-based (e.g., ID card) and knowledge-based (e.g., password) systems. However, the increased deployment of biometric systems in several commercial and government applications has raised questions about the security of the biometric system itself. Since the biometric traits of a user cannot be replaced if compromised, it is imperative that these systems are suitably secure in order to protect the privacy of the user as well as the integrity of the overall system. In this paper, we first investigate several methods that have been proposed in the literature to increase the security of the templates residing in a biometric system. We next propose a novel fingerprint matching architecture for resource-constrained devices (e.g., smart cards) that ensures the security of the minutiae templates present in the device. Experimental results describing the impact of several system parameters on the matching performance as well as the computational and storage costs are provided. The proposed architecture is shown to enhance the security of the minutiae templates while maintaining the overall matching performance of the system. [ABSTRACT FROM AUTHOR]

This thesis is on fingerprint recognition method and system, more in detail, the layered fingerprint recognition method and system to compare not only the minutiae or singular point, but also compare the images of detailed area for making prompt and accurate comparison of fingerprint. Since Edward R. Henry established the modern fingerprint approach in the recent days, the fingerprint has been applied in numerous fields. In particular, fingerprint is a powerful personal verification means that has been widely used for financial transactions, crime investigation and various security systems [1,2]. The contents proposed in this thesis calculate the relativity between the minutiae and extract certain area for the standard on similar patterns in measuring the similarity that it provides effective identification on the consistency on fingerprint with similar minutiae and fingerprint with prompt and rotation in fabric [9].First, the singular point and the minutiae are compared to seek the similar pattern promptly, then compare with the image of certain area based on the singular point (core point or delta point) that more accurate fingerprint can be recognized by performing the matching processes in multi-stages. Through the experiment that applies this theory, we have confirmed that there is a great improvement of verification rate and erroneous recognition rate of other person's fingerprint. [ABSTRACT FROM AUTHOR]

Automatic Fingerprint Authentication Systems are rapidly being incorporated in a wide range of applications, satisfying the society demand of accurate identification frameworks in order to prevent unauthorised accesses or fraudulent uses. Most of biometrics based personal identification systems run on high-performance computer based platforms, which execute a set of complex algorithms implemented in software. Those solutions cannot be applied to small, low-cost and low-power embedded systems, based on microprocessors without floating-point arithmetic unit. In this article we present a hardware/software implementation of a fingerprint minutiae extraction algorithm. The proposed system consists of a microprocessor and a coprocessor implemented in an associated FPGA. In order to develop an efficient implementation, fixed-point computations have substituted the floating-point ones. Due to the low feature requirements the whole system is suitable for a SoC with embedded flexible hardware. [ABSTRACT FROM AUTHOR]

This paper presents an efficient minutiae-based fingerprint representation and matching method using the distribution of distances between points. The proposed method uses the distribution of pairwise distances between minutiae as fingerprint features. The fingerprint matching between the input and the template fingerprints is performed by considering the Euclidean distance between the distributions. Most conventional minutiae matching methods require intensive comparing in order to align the two fingerprints translationally and rotationally, whereas the proposed method does not need such an intensive comparison procedure for the alignment. In addition, the feature vector generated by the proposed method has a small and fixed length, which is more advantageous in some applications such as smart cards. The experiments using the randomly generated 800 minutiae sets and our database consisting of 800 fingerprints show that the proposed method can be used effectively in applications that have limited memory and require high speed. [ABSTRACT FROM AUTHOR]

This paper proposes a new fingerprint matching algorithm for locally deformed fingerprints based on a geometric structure of minutiae, called the radial structure, which is a collection of lines from a minutia that connect its Voronoi neighbors. The proposed algorithm consists of local matching followed by global matching, in both of which a new robust scoring strategy is employed. The local matching compares individual radial structures of a query and a template, and the global matching, performed when the local matching fails, utilizes overall radial structures of a query. The algorithm has been tested using the FVC2002 DB1 fingerprint database on a Pentium-4 personal computer with 1.8 GHz clock and 256 Mbyte RAM. The test results show that the average matching time including preprocessing is 0.9 sec, and the equal error rate is 8.22%. It has been observed that the proposed algorithm has a smaller equal error rate by 7.18% than Mital and Teoh's. This is a substantial improvement in the equal error rate on the angle-distance based algorithm of Mital and Teoh. This improvement is attributed to the following features of the proposed algorithm: the radial structure is obtained from Voronoi neighboring minutiae, which results in more robustness to false minutiae; and the scoring strategy rewards similarity in the geometric structure rather than feature types as in Mital and Teoh's algorithm. [ABSTRACT FROM AUTHOR]

Coping with non-linear distortions in fingerprint matching is a real challenging task. This paper proposed a novel method, fuzzy features match (FFM), to match the deformed fingerprints. The fingerprint was represented by the fuzzy features: local triangle features set. The similarity between fuzzy features is used to character the similarity between fingerprints. First, a fuzzy similarity measure for two triangles was introduced. Second, the result is extended to construct a similarity vector which includes the triangle-level similarities for all triangles in two fingerprints. Accordingly, a similarity vector pair is defined to illustrate the similarities between two fingerprints. Finally, the FFM measure maps a similarity vector pair to a scalar quantity, within the real interval [0, 1], which quantifies the overall image to image similarity. To validate the method, fingerprints of FVC2004 were evaluated with the proposed algorithm. Experimental results show that FFM is a reliable and effective algorithm for fingerprint matching with non-liner distortions. [ABSTRACT FROM AUTHOR]