Showing papers in "International Journal of Biometrics in 2015"

Biometric signature verification

Abstract: In recent years, biometric signature verification BSV has been considered with renewed interest with increasing need of security and individual verification and authentication whether in banks, offices, institutions or other commercial organisations. Biometric signature verification is a behavioural biometric technique as a signature signifies unique behaviour of an individual. It can upgrade online banking using online digital systems for signing which cannot be altered or manipulated. Digital signature pads use algorithms to record the features of the signature, which is used to authenticate a signer during a transaction. This paper aims to present a comprehensive literature survey of the most recent research papers on biometric signature verification. It highlights the most important methods and addresses variations in the methods and features that are being taken up in the most recent research in this field along with the possible extensions.

Invariant face recognition using Zernike moments combined with feed forward neural network

Abstract: The paper proposes a face recognition system using Zernike moments ZM and feed forward neural network as a classifier. Magnitudes of the ZM, which are invariant to rotation, are used as feature vectors for efficient representation of the images. The experiment was conducted on the ORL and Texas 3D Face Recognition Database which has both colour and range images. The recognition performance with measures like overall recognition accuracy, false acceptance rate, false rejection rate and true rejection rate was evaluated with multilayer perceptron neural network, radial basis function neural network and probabilistic neural network for variable lengths of the feature vector using confusion matrix. The simulation results indicates that the invariant ZM with neural network classifier was successful in recognising the images constrained to different variations and illumination conditions. The overall classification accuracy of 99.7% with MLPNN and 99.6% with MLPNN was achieved with range images and grey images from Texas 3D Face Recognition Database, respectively. Furthermore, 99.5% accuracy with RBFNN was achieved from ORL database.

Approach to cryptographic key generation from fingerprint biometrics

Abstract: In crypto-biometric system CBS, biometric is combined with cryptography. In CBS, either accessing a cryptographic key is controlled with biometric or the key is generated from biometric features. This work is related to the latter approach in CBS. In such a system, protecting the privacy of the biometric data is an important concern. Further, there is a need to generate different cryptographic keys from the same biometric template of a user. Cancellable transformation of biometric data prior to the key generation is known as a solution. In this paper, we propose an approach to generate cryptographic key from cancellable fingerprint templates C

Local texture description framework-based modified local directional number pattern: a new descriptor for face recognition

Abstract: Texture descriptors effectively capture the surface property of images. However, almost all the available local texture descriptors encode a texture pattern using the closest neighbours. But the local texture description framework LTDF proposed earlier proved the importance of eight sampling points that lie elliptically at a certain distance apart from a pixel under consideration in distinguishing different face images. Recently, a local texture descriptor namely local directional number pattern LDN is introduced to encode the directional information of the structure of a face's texture. Incorporating the concepts of LTDF and LDN, this paper proposes a new texture descriptor namely LTDF-based modified local directional number pattern LTDF_MLDN. LTDF_MLDN describes a texture pattern with the sampling points at dissimilar area. Effectiveness of the system is tested for the different issues in face recognition using five benchmark databases. Experimental results reveal the effectiveness of the proposed descriptor over the state-of-the-art approaches.

Palmprint identification and verification based on wide principal lines through dynamic ROI

Abstract: In this paper, a novel palmprint identification and verification algorithm is proposed based on wide principal lines through dynamic ROI. Region of interest ROI extraction is an important task for palmprint identification. Earlier reported works used fixed size ROI for the recognition of palmprints. When the fixed size ROI is used the palm area taken up for recognition is less compared to dynamic ROI extraction. The proposed algorithm focuses on extraction of maximum possible ROI. A set of wide principal line extractors are devised. Later these wide principal line extractors are used to extract the wide principal lines from dynamic ROI. A two stage palmprint identification algorithm is proposed based on wide principal lines. The experimental results demonstrate that the proposed approach extracts better ROI on the PolyUPalmprint Database when compared to the existing fixed size and dynamic size ROI extraction techniques. The experimental results for the verification and identification on PolyUPalmprint Database show that the discrimination of wide principal lines is also strong.

Research of dual-modal decision level fusion for fingerprint and finger vein image

Abstract: The use of personal identity authentication systems with multi-modal biometrics has been proposed in order to increase the performance and robustness against environmental variation and fraudulent attacks. This paper presents a novel fingerprint and finger vein identity authentication system based on multi-route detection. Firstly, two classifiers are designed for fingerprint image and finger vein image respectively. Then extracted feature vectors from the first stage are then concatenated to make the third classifier. The final result is achieved by the fusion of the three classifiers' recognition results at the decision level. Experimental results show that this algorithm not only overcomes the limitations of single-modal biometrics, but also effectively improves the recognition performance of the system.

3D face recognition based on normal map features using selected Gabor filters and linear discriminant analysis

Abstract: In this paper, we present a new approach to enhance and improve the performance of automatic 3D face recognition system. The proposed method has been implemented through a preprocessing technique to align and normalise all images in the database based on eyes centres localisation using 2D normalised cross-correlation 2DNCC. Preprocessing 3D face data has been implemented using depth map representation of the 3D data. The detected eyes centres and eyes distance ED have been used to segment and align 3D face images to produce a cropped face region of interest ROI. The proposed approach extracted 3D face features using a set of selected orthogonal Gabor filters applied to normal map representation of the 3D face model. This approach minimises the feature vector extracted compared to systems that use complete Gabor filters bank. A further compression to the extracted features has been accomplished using linear discriminant analysis LDA before the classification stage. Experimental results show that the proposed system is effective for both dimensionality reduction and good recognition performance when compared to current systems. The system has been tested using CASIA and Gavab 3D face images databases and achieved 98.35% and 85% recognition rates, respectively.

Fast match-on-card technique using in-matcher clustering with ISO minutia template

Abstract: Fingerprint match-on-card is receiving more attention from society because of higher level of security and less privacy concern. The ISO/IEC19794-2 for match-on-card defines the minutiae data format that contains the basic information. Therefore, match-on-card is a challenge, especially when dealing with deformation and common correspondence for alignment. In this paper, a novel in-matcher clustering method is proposed to search for the matched minutia clusters to cater for deformation. Moreover, a further matching step using Mahalanobis distance to measure the inter-cluster similarity is proposed to remove the wrongly matched clusters. Finally, the overall match score is generated by combining the scores from matched clusters and the geometrical structure of clusters. The proposed algorithm achieved an average EER ⇐ 5.1979% using all FVC databases. In the NIST evaluation, the achieved false match rate FMR = 0.001 and false non-match rate FNMR = 0.08 and the average on-card verification time is 1.01s using 8-bit smartcard.

Recognition accuracy of the new fingerprint construct based on a compact minutiae pattern

Abstract: Instead of using the entire minutiae template to generate a protected fingerprint template, recently a non-invertible cancellable fingerprint construct based on a 3-5 minutiae pattern was proposed as a safer alternative. This paper investigates the recognition accuracy attainable by this new fingerprint construct. It is found that using five samples of a person's reference fingerprint and allowing for a maximum of three authentication attempts provides a genuine user with the best chance of being successfully authenticated. An evaluation of the FAR and FRR in this scenario demonstrates that the new fingerprint construct can be tuned to suit the performance and security requirements of different applications by adjusting the pattern size and matching thresholds. The fingerprint construct is then modified to improve its ability to discriminate between genuine users and impostors. Compared to other non-invertible fingerprint template protection schemes, the performance of the modified fingerprint construct is found to be favourable.

Face verification using local binary patterns and generic model adaptation

Abstract: The popular local binary patterns LBP have been highly successful in representing and recognising faces. However, the original LBP-based face recognition method has some problems that need to be addressed. In this work, we propose two approaches to address the histogram representation drawbacks in the LBP-based face verification system. The first approach employs vector quantisation maximum a posteriori adaptation VQMAP model, where a generic face model is obtained by vector quantisation and the user models are inferred using maximum a posteriori adaptation. The second approach proposes an enhanced LBP histogram representation by adapting a generic face histogram to each user. Moreover, the two proposed approaches are further fused to enhance the verification performance. We evaluate our proposed approaches on two publicly available databases, namely BANCA and XM2VTS, and compare the results against the original LBP approach and its variants, demonstrating very promising results.

Face recognition using a novel image representation scheme and multi-scale local features

Abstract: This paper presents a new method for improving face recognition performance under difficult conditions. Specifically, a new image representation scheme is proposed which is derived from the YCrQ colour space using principal component analysis PCA followed by Fisher linear discriminant analysis FLDA. A multi-scale local feature, LBP-DWT, is used for face representation which is computed by extracting different resolution local binary patterns LBP features from the new image representation and transforming the LBP features into the wavelet domain using discrete wavelet transform DWT and Haar wavelets. A variant of non-parametric discriminant analysis NDA, called regularised non-parametric discriminant analysis RNDA is introduced to extract the most discriminating features from LBP-DWT. The proposed methodology has been evaluated using two challenging face databases FERET and multi-PIE. The promising experimental results show that the proposed method outperforms two state-of-the-art methods, one based on Gabor features and the other based on sparse representation classification SRC.

Fingerprint template protection scheme based on partial minutiae patterns: a comprehensive non-invertibility analysis

Abstract: A crucial property of an effective fingerprint template protection scheme is non-invertibility, which ensures that the original fingerprint template cannot be recovered from its secured counterpart. Since it is extremely difficult to design a function that achieves a high degree of non-invertibility, it is unsafe to use an entire fingerprint template as the input to any such function. Most existing fingerprint template protection mechanisms, however, do exactly this. One scheme that stands apart from the rest, in that its protected template originates from only a small portion of the minutiae template, is a new fingerprint construct based on partial minutiae patterns. This paper presents a comprehensive analysis of the non-invertibility of this scheme. It is found that the new fingerprint construct has extremely strong non-invertibility and good resistance to a Record Multiplicity Attack in practice. We also propose a modification to improve the method's resistance to this type of attack.

Hough transform-based technique for fingerprint alteration detection and classification

Abstract: Fingerprint alteration is a major threat to automatic fingerprint identification systems, especially in border security control system. A Hough transform-based method for detection and classification of altered fingerprint is presented here. Altered fingerprints consists of huge amount of broken ridges due to different process used for making alteration and this in turn causes large number of ridge ending. The amount of ridge endings is different in different types of altered fingerprints. Hough transform-based method proposed in this paper utilises the variation in ridge ending density as a feature for detection and classification of altered fingerprints. The ridge end points in normal and altered fingerprints are collinear even though they are distributed randomly in the image space. Due to the variation of ridge ending density, the number of collinear ridge end points varies with respect to normal and different types of alteration. Making use of this, a threshold is selected in the Hough accumulator to perform detection and classification of fingerprint alteration. A method for the classification of scar present in altered and normal fingerprints is also proposed here.

Implementation of modified polar complex moments-based fingerprint orientation estimation for effective segmentation

Abstract: Segmentation is an important step in deciding the performance of fingerprint identification systems. In this paper, we present the modified polar complex moments MPCMs fingerprint orientation estimation algorithm, capable of describing the fingerprint flow structures including singular point regions in the fingerprint images effectively. To discard the background region of the low-quality fingerprint images, regularisation was employed. These algorithms are tested on various types of fingerprint images containing low-quality unrecoverable region and the results obtained from the proposed method were compared with those obtained from well-known gradient-based and PCMs methods. The proposed method was also used to study the contrast enhancement process with our previously developed modified histogram equalisation MHE based on adaptive inverse hyperbolic tangent AIHT method. The MPCMs method exhibits better segmentation than the traditional methods in both normal and low-contrast image segmentations, as evident from the estimated matching scores as well as ROC graph.

Gender recognition from biologically guided anthropometric features

Abstract: Automated gender recognition from whole body images is a challenging problem with multi-disciplinary utility. A greater understanding of potential feature components e.g., anthropometry, movement, etc. may help future feature selection algorithms better target effective features, reduce feature complexity, and increase algorithm generalisability. In this study we evaluated the potential of static anthropometric measurements for gender recognition. Utilising a large 3D body scan repository, we first captured novel measurements directly relevant to computer vision applications, and used these to create biologically guided feature sets. Linear discriminant analysis was used to classify gender across specific demographics to additionally evaluate the potentially confounding influences of race, age, and obesity. The effects of view angle were also preliminarily analysed. Classification results showed greater accuracy in the frontal plane than the sagittal plane, with models reaching 99% and 96% accuracy, respectively. Feature rankings and correlations are presented and discussed in relevance to future algorithms.

Efficient palmprint identification using novel symmetry filter and alignment refinement

Abstract: This paper presents a robust algorithm for line orientation code-based palmprint identification in which we propose a novel symmetry filter and an efficient alignment refinement technique. The main idea of the symmetry filter is to compute the approximate magnitude of the Gabor filter based on the modified finite Radon transform MFRAT, the so-called GMFRAT filter. The advantages of GMFRAT filters are that: 1 they are capable of quickly computing orientation codes; 2 they remarkably reduce remarkably the sizes of these features. The alignment refinement technique, which uses local orientation patterns, is also proposed to solve the problem of rotations and translations caused by an imperfect preprocessing phase. Based on our alignment refinement, the matching algorithm is designed. The experimental results obtained using the public databases of the Hong Kong Polytechnic University and the Indian Institute of Technology Delhi demonstrate the effectiveness of the proposed method.

An embedded system for extracting keystroke patterns using pressure sensors

Abstract: Popular biometric security technologies include fingerprint and iris recognition systems. These technologies are extremely accurate because the patterns associated with an individual's finger or eye are very unique and static. However, when these technologies are used for physical access control they inform the potential adversary that specific characteristics are required to gain access. Behaviometrics aims to develop new strategies to enhance physical security via covert monitoring of distinct behavioral patterns. This research presents a novel stand-alone behaviometric prototype that incorporates standard password security with unique pressure characteristics to covertly analyse individual typing patterns. The prototype is evaluated under a controlled setting with 62 human subjects and nine classification algorithms. The kNN algorithm produced the highest classification rate of 94%. This research is one of the few papers that empirically substantiates biometric performance with a large-scale human subject trial, and also identifies several critical design considerations that impact classifier performance.

Face recognition based on 2D and 3D data fusion

Abstract: The aim of the work presented in this paper is to present current state of the art of face recognition methods and describe proposal algorithms for face biometric identification that analyse 2D face images and 3D face geometry scans. Data for analysis gathered via 3D scanner are processed through different phases. These are: segmentation phase, feature extraction phase and comparison phase. Segmentation relies on localising characteristic landmark points of the face and projecting the face point cloud onto a plane constructed on the basis of these characteristic points. Feature extraction phase calculates separate feature vectors for 2D and 3D input data. Comparison phase applies fusion of 2D and 3D methods and calculates similarity value between two samples. All samples are compared against one another and results presented as DET curves are generated. By analysis of DET curves, conclusions are formulated.

Phoneme dependent inter-session variability reduction for speaker verification

Abstract: GMM-UBM super-vectors will potentially lead to worse modelling for speaker verification due to the inter-session variability, especially when a small amount of training utterances were available. In this study, we propose a phoneme dependent method to suppress the inter-session variability. A speaker's model can be represented by several various phoneme Gaussian mixture models. Each of them covers an individual phoneme whose inter-session variability can be constrained in an inter-session independent subspace constructed by principal component analysis PCA, and it uses corpus uttered by a single speaker that has been recorded over a long period. SVM-based experiments performed using a large corpus, constructed by the National Research Institute of Police Science NRIPS to evaluate Japanese speaker recognition, and demonstrate the improvements gained from the proposed method.