Search

Abstract: Unimodal analysis of palmprint and palm vein has been investigated for person recognition. One of the problems with unimodality is that the unimodal biometric is less accurate and vulnerable to spoofing, as the data can be imitated or forged. In this paper, we present a multimodal personal identification system using palmprint and palm vein images with their fusion applied at the image level. The palmprint and palm vein images are fused by a new edge-preserving and contrast-enhancing wavelet fusion method in which the modified multiscale edges of the palmprint and palm vein images are combined. We developed a fusion rule that enhances the discriminatory information in the images. Here, a novel palm representation, called “Laplacianpalm” feature, is extracted from the fused images by the locality preserving projections (LPP). Unlike the Eigenpalm approach, the “Laplacianpalm” finds an embedding that preserves local information and yields a palm space that best detects the essential manifold structure. We compare the proposed “Laplacianpalm” approach with the Fisherpalm and Eigenpalm methods on a large data set. Experimental results show that the proposed “Laplacianpalm” approach provides a better representation and achieves lower error rates in palm recognition. Furthermore, the proposed multimodal method outperforms any of its individual modality. [Copyright &y& Elsevier]

Abstract: Fingerprint classification is crucial to reduce the processing time in a large-scale database. Two popular features used are the singularities and orientation information and they are complementary. Therefore, an algorithm based on the interactive validation of singular points and the constrained nonlinear orientation model is proposed. The final features used for classification comprises the coefficients of the orientation model and the singularity information. This resulted in very compact feature vector which is used as input to an SVM classifier to perform the classification. The experiments conducted on the NIST database 4 show the effectiveness of the proposed method in producing good classification result. [Copyright &y& Elsevier]

Abstract: In this paper, we developed an algorithm to model the fingerprint orientation field. The algorithm comprises two steps, orientation prediction and model computation. Orientation prediction is based on piece-wise first-order phase portrait model. It is used to estimate the orientation in areas where there is no ridge information in the input image or the coherence of the orientation field is low. In the model computation, a constrained nonlinear phase portrait algorithm is proposed, which aims to get an accurate mathematical model of the fingerprint orientation field. Compared to the prior works, this algorithm is able to predict orientation even in noisy regions and it integrates the global and local orientation description into a unified mathematical form. Experiments conducted on the first 500 images of the NIST-4 database showed that the proposed algorithm is able to model all the fingerprint orientation patterns except those with very poor quality images where the orientation information cannot be clearly extracted. [Copyright &y& Elsevier]

Abstract: Orientation pattern is an important feature for characterizing fingerprint and plays critical roles in fingerprint recognition and fingerprint classification. This paper proposes a framework for modeling the fingerprint orientation field based on the variational principle, where the orientation pattern can be estimated through solving the associated Euler–Lagrange equation. Compared with existing methods, our proposed method has the following features. Firstly, it does not require any prior information about the structure of the acquired fingerprint, such as location of singular point(s). Secondly, it explicitly provides freedom for modeling the singularity in the orientation field. Thirdly, it has less number of parameters. Comparison has been made with respect to state-of-the-arts in fingerprint orientation modeling in terms of modeling accuracy, fingerprint enhancement and singular point detection. Advantages of the proposed method are demonstrated. [Copyright &y& Elsevier]

Abstract: This paper presents a novel method for fingerprint orientation modeling, which executes in two phases. Firstly, the orientation field is reconstructed using a lower order Legendre polynomial to capture the global orientation pattern in the fingerprint structure. Then the preliminary model around the region with presence of fingerprint singularities is dynamically refined using a higher order Legendre polynomial. The singular region is automatically detected through the analysis on the orientation residual field between the original orientation field and the orientation model. The method does not require any prior knowledge on the fingerprint structure. To validate the performance, the method has been applied to fingerprint image enhancement, fingerprint singularity detection and fingerprint recognition using the FVC 2004 data sets. Compared with the recently published Legendre polynomial model, the proposed method attains higher accuracy in fingerprint singularity detection, lower error rates in fingerprint matching. [ABSTRACT FROM AUTHOR]

Abstract: This work addresses the problem of deformable hand shape recognition in biometric systems without any positioning aids. We separate and recognize multiple rigid fingers under Euclidean transformations. An elliptical model is introduced to represent fingers and accelerate the search of optimal alignments of fingers. Unlike other methods, the similarity is measured during alignment search based on finger width measurements defined at nodes by controllable intervals to achieve balanceable recognition accuracy and computational cost. Technically, our method bridges the traditional handcrafted-feature methods and the shape-distance methods. We have tested it using our 108-person-540-sample database with significantly increased positive recognition accuracy. [Copyright &y& Elsevier]

Subspace learning aims to learn a projection matrix from a given training set so that a transformation of raw data to a low-dimensional representation can be obtained. In practice, the labels of some training samples are available, which can be used to improve the discrimination of low-dimensional representation. In this paper, we propose a semi-supervised learning method which is inspired by the biological observation of similar inputs having similar codes (SISC), i.e., the same collection of cortical columns of the mammal׳s visual cortex is always activated by the similar stimuli. More specifically, we propose a mathematical formulation of SISC which minimizes the distance among the data points with the same label while maximizing the separability between different subjects in the projection space. The proposed method, namely, semi-supervised L2graph (SeL2graph) has two advantages: (1) unlike the classical dimension reduction methods such as principle component analysis, SeL2graph can automatically determine the dimension of feature space. This remarkably reduces the effort to find an optimal feature dimension for a good performance; and (2) it fully exploits the prior knowledge carried by the labeled samples and thus the obtained features are with higher discrimination and compactness. Extensive experiments show that the proposed method outperforms 7 subspace learning algorithms on 15 data sets with respect to classification accuracy, computational efficiency, and robustness to noises and disguises. [ABSTRACT FROM AUTHOR]

Abstract: The area under the ROC curve (AUC) provides a good scalar measure of ranking performance without requiring a specific threshold for performance comparison among classifiers. AUC is useful for imprecise environments since it operates independently with respect to class distributions and misclassification costs. A direct optimization of this AUC criterion thus becomes a natural choice for binary classifier design. However, a direct formulation based on the AUC criterion would require a high computational cost due to the drastically increasing input pair features. In this paper, we propose an online learning algorithm to circumvent this computational problem for binary classification. Different from those conventional recursive formulations, the proposed formulation involves a pairwise cost function which pairs up a newly arrived data point with those of opposite class in stored data. Moreover, with incorporation of a sparse learning into the online formulation, the computational effort can be significantly reduced. Our empirical results on three different scales of public databases show promising potential in terms of classification AUC, accuracy, and computational efficiency. [Copyright &y& Elsevier]