Showing papers on "Signature recognition published in 2002"

Diagnostic systems and methods for predictive condition monitoring

Abstract: A system for empirically diagnosing a condition of a monitored system. Estimates of monitored parameters from a model of the system provide residual values that can be analyzed for failure mode signature recognition. Residual values can also be tested for alert (non-zero) conditions, and patterns of alerts thus generated are analyzed for failure mode signature patterns. The system employs a similarity operator for signature recognition and also for parameter estimation. Failure modes are empirically determined, and precursor data is automatically analyzed to determine differentiable signatures for failure modes.

Biometric hash based on statistical features of online signatures

Abstract: Presents an approach to generating biometric hash values based on statistical features in online signature signals. Whilst the output of typical online signature verification systems are threshold-based true-false decisions, based on a comparison between test sample signals and sets of reference signals, our system responds to a signature input with a biometric hash vector, which is calculated based on an individual interval matrix. Especially for applications, which require key management strategies, hash values are of great interest, as keys can be derived directly from the hash value, whereas a verification decision can only grant or refuse access to a stored key. Further, our approach does not require storage of templates for reference signatures, thus increases the security of the system. In our prototype implementation, the generated biometric hash values are calculated on a pen-based PDA and used for key generation for a future secure data communication between a PDA and a server by encryption. First tests show that the system is actuality able to generate stable biometric hash values of the users and although the system was exposed to skilled forgeries, no test person was able to reproduce another subject's hash vector.

Image Recognition and Classification: Algorithms, Systems, and Applications

Abstract: Introduction linear filters for pattern recognition nonlinear filtering for image recognition distortion invariant pattern recognition systems image recognition based on statistical detection theory neural networks based automatic target recognition hyperspectral automatic object recognition laser radar automatic target recognition radar signature recognition wavelets for image recognition pattern recognition for anticounterfeiting and security systems applications of pattern recognition techniques to road sign recognition and tracking optical and optoelectronic implementation of linear and nonlinear filters.

Spatial frequency domain image processing for biometric recognition

Abstract: Biometric recognition refers to the process of matching an input biometric to stored biometric information. In particular, biometric verification refers to matching the live biometric input from an individual to the stored biometric template about that individual. Examples of biometrics include face images, fingerprint images, iris images, retinal scans, etc. Thus, image processing techniques prove useful in the biometric recognition. We discuss spatial frequency domain image processing methods useful for biometric recognition.

Recognition of off-line handwritten Arabic words using hidden Markov model approach

Abstract: Hidden Markov models (HMM) have been used with some success in recognizing printed Arabic words. In this paper, a complete scheme for totally unconstrained Arabic handwritten word recognition based on a model discriminant HMM is presented. A complete system able to classify Arabic handwritten words of one hundred different writers is proposed and discussed. The system first attempts to remove some of variation in the images that do not affect the identity of the handwritten word. Next, the system codes the skeleton and edge of the word so that feature information about the lines in the skeleton is extracted. Then a classification process based on the HMM approach is used. The output is a word in the dictionary. A detailed experiment is carried out and successful recognition results are reported.

A multi-sample multi-source model for biometric authentication

Abstract: In this study, two techniques that can improve the authentication process are examined: (i) multiple samples and (ii) multiple biometric sources. We propose the fusion of multiple samples obtained from multiple biometric sources at the score level. By using the average operator, both the theoretical and empirical results show that integrating as many samples and as many biometric sources as possible can improve the overall reliability of the system. This strategy is called the multi-sample multi-source approach. This strategy was tested on a real-life database using neural networks trained in one-versus-all configuration.

Audio-visual continuous speech recognition using a coupled hidden Markov model.

Abstract: With the increase in the computational complexity of recent computers, audio-visual speech recognition (AVSR) became an attractive research topic that can lead to a robust solution for speech recognition in noisy environments. In the audio visual continuous speech recognition system presented in this paper, the audio and visual observation sequences are integrated using a coupled hidden Markov model (CHMM). The statistical properties of the CHMM can describe the asyncrony of the audio and visual features while preserving their natural correlation over time. The experimental results show that the current system tested on the XM2VTS database reduces the error rate of the audio only speech recognition system at SNR of 0db by over 55%.

Analysis of stability in hand-written dynamic signatures

Abstract: This paper presents a new technique to evaluate the local stability in hand-written dynamic signatures and use the results to improve the process of automated signature verification. The experimental results points out the usefulness of integrating stability, information into the signature verification process.

Development of Isolated Word Speech Recognition System

Abstract: The isolated word speech recognition system based on dynamic time warping (DTW) has been developed. Speaker adaptation is performed using speaker recognition techniques. Vector quantization is used to create reference templates for speaker recognition. Linear predictive coding (LPC) parameters are used as features for recognition. Performance is evaluated using 12 words of Lithuanian language pronounced ten times by ten speakers.

From Biometrics Technology to Applications Regarding Face, Voice, Signature and Fingerprint Recognition Systems

Abstract: In this chapter, several biometric recognition systems, based on voice, fingerprint, face and signature are presented. The description of the state-of-the-art technologies regarding these biometric characteristics is widely faced. Minutiae extraction-based fingerprint matching, GMM-based speaker verification, on-line HMM-based signature verification, and PCA- or LDA-based face recognition are quoted. We will also focus on multimodality and data fusion in biometric systems; finally, some application strategies and some real-world demos are described.

Grid- and dummy-cluster-based learning of normal and intrusive clusters for computer intrusion detection

Abstract: As an important part of information security, computer intrusion detection aims at capturing intrusive activities occurring in computer and network systems. Many existing signature recognition techniques for intrusion detection cannot handle huge amounts of complex data from computer and network systems to detect intrusions in a scalable, incremental manner. This paper presents an application of an innovative data-mining algorithm—CCAS—to intrusion detection through intrusion signature recognition. CCAS provides a scalable, incremental procedure to learn clusters of different classes (i.e. normal and intrusive classes) from historic training data of normal and intrusive activities in computer and network systems. These clusters of normal and intrusive computer activities are used to classify observed data of computer activities for intrusion detection. Two different methods of learning clusters are developed, tested and compared: grid based and dummy-cluster based. Training and testing data are computer audit data produced by the Basic Security Module of a Solaris operating system to record activities in a UNIX-based host machine. The two methods of CCAS are tested using four different input orders of training data points to examine the robustness (sensitivity) of these methods to the input order of training data points. The detection performance and robustness of both CCAS methods are analyzed. The testing results show that different input orders of training data have a certain impact on the performance of both methods. The impact on the performance of CCAS based on dummy clusters is more significant when all normal data are presented first, before attack data in the training data set. CCAS based on dummy clusters produces a better performance than grid-based CCAS for three of the four input orders, and in overall produces fewer clusters and thus requires less computation time in clustering and classification for intrusion detection. Copyright © 2002 John Wiley & Sons, Ltd.

A system for fingerprint minutiae classification and recognition

Abstract: This paper shows a study about biometrics characteristics for recognition/classification and presents how it is used for individual recognition. The approach uses the automated fingerprint recognition based on minutia, which are extracted directly from the finger prints and the methodology used to its recognition is the artificial neural networks (ANN) based system. The neocognitron model was the ANN chosen. Inasmuch as neocognitron was originally implemented for handwritten characters recognition, it is possible to verify its usefulness for another kind of pattern recognition. Finally it is presented the results for this system and the conclusions according to the number of samples and recognition rate.

DVHMM: variable length text recognition error model

Abstract: This paper proposes a text recognition error model called the dual variable length output hidden Markov model (DVHMM) and gives a parameter estimation algorithm based on the EM algorithm. Although existing probabilistic error models are limited to substitution (1, 1), insertion (1, 0), and deletion (0, 1) errors, the DVHMM can handle error patterns of any pair (i, j) of lengths including substitution, insertion, and deletion.

Facial expression recognition using pseudo 3-D hidden Markov models

Abstract: In this paper, pseudo 3-D hidden Markov models (P3DHMMs) are applied to the task of dynamic facial expression recognition. P3DHMMs are an extension of the pseudo 2-D case, which has been successfully used for the classification of images and the recognition of faces. Although the application of P3DHMMs for image sequence recognition has been reported before, this paper provides a formal definition of the novel approach as well as a detailed explanation of a triple embedded Viterbi algorithm. Furthermore, an equivalent one-dimensional structure is introduced, which allows the application of the standard Viterbi and Baum-Welch algorithms. The approach has been evaluated on a person independent database, which consists of 4 different facial expressions performed by 6 individuals. The recognition accuracy achieved in the experiments is close to 90%.

Feature selection for off-line recognition of different size signatures

Abstract: The aim of this work is to select a set of features, which have good performance to solve the problem of signature recognition of different sizes. The signature database was formed for three sizes of signatures per user, small, median and big. This study uses structural features, pseudo-dynamic features and five moment groups. The feature selection method chosen is the one that select the best individual features based on classifiers like Bayes and k-NN.

Biometric applications based on handwriting

Abstract: A wide variety of biometric based techniques have been proposed but it is quite difficult to classify the approaches according to their application domains and to measure their functionality. Our intention is to classify today's applications in detail for one particular biometric scheme, handwriting. To give individual users with a specific application in mind orientation and a decision tool, we have built a new classification scheme and furthermore define major characteristics for each of the application classes as an evaluation matrix.

Memory space reduction for hidden Markov models in low-resource speech recognition systems.

Abstract: Low-cost recognition systems based on hidden Markov models (HMM) for mobile speech recognizers (mobile phones, PDAs) have a limited quantity of memory and processing power. Furthermore, the resources have to be shared between several applications. In this paper memory efficient HMMs were investigated for low-cost recognition platforms. The feature parameter tying HMM and subspace distribution clustering HMM (SDCHMM) were explored. In order to achieve less memory requirements, a shared codebook approach for feature parameter tying HMM and SDCHMM was developed and its effectiveness was experimentally proved. It was shown that this approach leads to a relative increase of word error rate of less than 10% for 50% of memory reduction.

Combining HMM-based two-pass classifiers for off-line word recognition

Abstract: For off-line recognition of cursive handwritten word, the intersection between segmentation and recognition is complicated and makes the recognition problem still a challenging task Hidden Markov models (HMMs) have the ability to perform segmentation and recognition in a single step In this paper we present an HMM based unsymmetric two-pass modeling approach for recognizing cursive handwritten word The two-pass recognition approach exploits the segmentation ability of the Viterbi algorithm and creates three different HMM sets and carries out two passes of recognition A weighted voting approach is used to combine results of the two recognition passes A high recognition rate was achieved for recognizing cursive handwritten words with a lexicon of 1120 words An experiment on NIST sample hand print data of ten different writers was also carried out The experimental results demonstrate that the two-pass approach can achieve better recognition performance and reduce the relative error rate significantly

Dynamic biometric person authentication using pen signature trajectories

Abstract: Authentication of individuals is rapidly becoming an important issue. The authors have previously proposed a pen-input on-line signature verification algorithm incorporating pen-position, pen-pressure and pen-inclination trajectories. This paper proposes an algorithm with three new features: (i) Incorporation of pen-velocity trajectories, (ii) A new distance measure between two signatures, and (iii) A new efficient algorithm for computing the distance measure. Preliminary experimental result looks encouraging.

HMM based online hand-drawn graphic symbol recognition

Abstract: In this paper, an online hand-drawn graphic symbol recognition algorithm based on hidden Markov models is presented. A rearrangement strategy is applied to the hand-drawn symbol points in order to alleviate the influence of the difference in drawing sequence. Based on rearranged drawing points, global distance measure and local angle feature are extracted as the feature vector. After the quantization, a discrete HMM is used as the core recognizer. The experiment shows the recognition rate of our system can be above 85%.

A Neural-Network-Based Approach to Optical Symbol Recognition

Abstract: In this paper we propose a neural-network-based approach to solving optical symbol recognition problems, from node head recognition to handwritten digit recognition. We demonstrated that node heads could be easily recognized by using a set of fuzzy rules extracted from the parameters of trained neural networks. For handwritten digit recognition we demonstrated that only 12 features are sufficient to achieve a high recognition rate. Several databases were tested to demonstrate the effectiveness and efficiency of the proposed recognition method.

Fuzzy-logic-based sensor fusion of images

Abstract: The fusion of visual and infrared sensor images of potential driving hazards in static infrared and visual scenes is computed using the Fuzzy Logic Approach (FLA). The FLA is presented as a new method for combining images from different sensors for achieving an image that displays more information than either image separately. Fuzzy logic is a modeling approach that encodes expert knowledge directly and easily using rules. With the help of membership functions designed for the data set under study, the FLA can model and interpolate to enhance the contrast of the imagery. The Mamdani model is used to combine the images. The fused sensor images are compared to metrics to measure the increased perception of a driving hazard in the sensor-fused image. The metrics are correlated to experimental ranking of the image quality. A data set containing IR and visual images of driving hazards under different types of atmospheric contrast conditions is fused using the Fuzzy Logic Approach (FLA). A holographic matched-filter method (HMFM) is used to scan some of the more difficult images for automated detection. The image rankings are obtained by presenting imagery in the TARDEC Visual Perception Lab (VPL) to subjects. Probability of detection of a driving hazard is computed using data obtained in observer tests. The matched-filter is implemented for driving hazard recognition with a spatial filter designed to emulate holographic methods. One of the possible automatic target recognition devices implements digital/optical cross-correlator that would process sensor-fused images of targets. Such a device may be useful for enhanced automotive vision or military signature recognition of camouflaged vehicles. A textured clutter metric is compared to experimental rankings.

Online handwritten English word recognition based on cascade connection of character HMMs

Abstract: In this paper, a cascade connection hidden Markov model (CCHMM) method for online English word recognition is proposed. This model, which allows state transition, skip and duration, extends the way of HMM pattern description of handwriting English words. According to the statistic probabilities, the behavior of handwriting curve may be depicted more precisely. The Viterbi algorithm for the cascade connection model may be applied after the whole sample series of a word is input. Compared with the method of creating models for each word in lexicon, this method gives a faster recognition speed. Experiments show that CCHMM approach could obtain 89.26% recognition rate for the first candidate, while the combination of character and ligature HMM method's first candidate is 82.34%.

A tied-mixture 2D HMM face recognition system

Abstract: In this paper, a simplified 2D second-order hidden Markov model (HMM) with tied state mixtures is applied to the face recognition problem. The mixture of the model states is fully-tied across all models for lower complexity. Tying HMM parameters is a well-known solution for the problem of insufficient training data leading to nonrobust estimation. We show that parameter tying in HMM also enhances the resolution in the case of small model. The performance of the proposed 2D HMM tied-mixture system is studied for the face recognition problem and the expected improved robustness is confirmed.

On-line signature verification based on stroke matching

Abstract: Stability and style-variation are important characteristics in handwriting analysis and recognition. This paper describes a stability modelling technique of handwritings in the context of on-line signature verification. With this technique, the stability and style-variation characteristics are deduced from the dynamic warping relationship between the sequences of basic handwriting strokes of the reference samples. Reliablity measures of the extracted signature features are incorporated into the signature segmentation, model-building, and verification algorithms, so that stable handwriting features are emphasized in the signature matching process, while style-variations for less stable features are intentionally tolerated. The generated signature model consists of a structure description graph of the handwriting components and their stability information. A signature is accepted by the model if it is close to a permissible path within the weighted graph. A novel feature of the technique is its ability to refine the correspondence relation of the handwriting during model-building and signature verification. This ability enables the verification and partial correction of segmentation errors to reduce the number of false rejections while maintaining the system security level and keeping the number of reference samples manageable.

Tracking of feature and stroke positions for off-line signature verification

Abstract: There are inevitable variations in the signature patterns written by the same person. The variations can occur in the shape or in the relative positions of the characteristic features. For the set of training signature samples, two approaches are proposed. One approach measures the positional variations of the one-dimension projection profiles of the signature patterns, while the other determines the statistical variations in relative stroke positions of the two-dimensional signature patterns. Given a signature to be verified, the positional displacements are determined and the authenticity is decided based on the statistics of the training samples. A matrix estimation technique is also proposed to obtain a better estimation of the covariance matrix for dissimilarity computation. Results show that the proposed systems compare favorably with other methods.