Showing papers on "Signature recognition published in 2001"

Dynamic Time-Alignment Kernel in Support Vector Machine

Abstract: A new class of Support Vector Machine (SVM) that is applicable to sequential-pattern recognition such as speech recognition is developed by incorporating an idea of non-linear time alignment into the kernel function. Since the time-alignment operation of sequential pattern is embedded in the new kernel function, standard SVM training and classification algorithms can be employed without further modifications. The proposed SVM (DTAK-SVM) is evaluated in speaker-dependent speech recognition experiments of hand-segmented phoneme recognition. Preliminary experimental results show comparable recognition performance with hidden Markov models (HMMs).

Off-line signature verification using HMM for random, simple and skilled forgeries

Abstract: The problem of signature verification is in theory a pattern recognition task used to discriminate two classes, original and forgery signatures. Even after many efforts in order to develop new verification techniques for static signature verification, the influence of the forgery types has not been extensively studied. This paper reports the contribution to signature verification considering different forgery types in an HMM framework. The experiments have shown that the error rates of the simple and random forgery signatures are very closed. This reflects the real applications in which the simple forgeries represent the principal fraudulent case. In addition, the experiments show promising results in skilled forgery verification by using simple static and pseudodynamic features.

Complex signal decomposition and modeling

Abstract: A system (100), method and program product for monitoring a complex signal for ultrasensitive detection of state changes, or for signature recognition and classification is provided. A complex signal is decomposed periodically for empirical modeling (106). Wavelet analysis, frequency band filtering or other methods (102) may be used to decompose the complex signal into components. A library (104) of signature data may be referenced for selection of a recognized signature in the decomposed complex signal. The recognized signature may indicate data being carried in the complex signal. Estimated signal data may be generated for determination of an operational state of a monitored process or machine using a statistical hypothesis test (112) with reference to the decomposed input signal.

System for and method of web signature recognition system based on object map

Abstract: The present invention includes a system and a method of validating an identity of a user using a pointing device by comparing a sampled mouse signature with an authenticated mouse signature. The method includes presenting a virtual pad including a background graphic image, a plurality of objects positioned on the background graphic image to a user. The user moves the pointing device to manipulate a cursor on the background graphic image. The method includes a step of sampling a plurality of events corresponding to positions of the cursor to provide a sampled mouse signature including a set of position vectors. The present invention includes comparing the sampled mouse signature to a stored mouse signature representing the identity of a user, and validating the identity of a user in response to the comparing step.

Biometric identification method and system

Abstract: A biometric identification system ensuring reliable and protective identification of individuals even in a system having a biometric input device and a biometric verifier are separately provided is disclosed. The biometric data input device has a biometric data sensor and an encoder that encodes digital biometric data using secret information identifying the biometric data input device to transmit encoded data to the biometric verifier. The biometric verifier decodes the encoded data using the secret information to reproduce digital biometric data. The identity of the individual is verified when the digital biometric data is a registered biometric feature of an authorized user and the biometric data input device is an authorized device.

Novel biometric digital signatures for Internet‐based applications

Abstract: Personal identification numbers, passwords, smart cards and digital certificates are some of the means employed for user authentication in various electronic commerce applications. However, these means do not really identify a person, but only knowledge of some data or belonging of some determined object. This paper introduces the notion of biometric signature – a new approach to integrate biometrics with public key infrastructure, using biometric based digital signature generation which is secure, efficacious, fast, convenient, non‐invasive and correctly identifies the maker of a transaction. It also suggests two schemes for biometric signature using two existing and widely used digital signature algorithms, RSA and DSA, and discusses the problems associated with them individually. Speed of both schemes (based on iris recognition) is measured and compared with the help of JAVA implementation for both approaches.

System and method for determining likely identity in a biometric database

Abstract: A system and method for finding one or more target biometric samples that are similar to or match a query biometric sample. A query feature vector is generated from a query biometric vector. The query biometric vector represents the query biometric sample as a set of characteristics. The characteristics are either invariable or variable. The query feature vector comprises a plurality of features which are derived from the query biometric vector using a process that includes canonicalization of the characters in the biometric vector. The query feature vector is compared to a plurality of similarly created target feature vectors, each target feature vector representing a respective target biometric sample. A target biometric sample is a potential match to the query biometric sample when a threshold number of features in the corresponding target feature vector are identical to features in the query biometric vector.

Tree pattern system and method for multiple virus signature recognition

Abstract: A system, method and computer program product are provided for recognizing virus signatures. Initially, a list of virus signatures is provided. Next, the list of virus signatures is combined into a tree of virus signatures. Data is subsequently compared against the tree of virus signatures for virus signature recognition.

Recognition of human signatures

Abstract: We used a digitizing tablet to collect handwritten signatures, with five quantities recorded, namely horizontal and vertical pen tip position, pen tip pressure, and pen azimuth and altitude angles. We divided the signature features into visible ones, namely those related to an "image on the paper" and hidden ones, i.e. those using time-related observations. Cluster analysis was applied to segment the feature space into sub-regions of "similar" signatures. The classification function was approximated with the use of neural networks, namely a two-layer sigmoidal perceptron and the RCE network which is a variety of radial-basis network. Both signature classification and signature verification problems are considered.

Biometric Authentication Technology: From the Movies to Your Desktop

Abstract: A door silently opens, activated by a video camera and a face recognition system. Computer access is granted by checking a fingerprint. Access to a security vault is allowed after an iris check. Are these scenes from the latest Hollywood spy thriller? Perhaps, but soon it could be in your office or on your desktop. Biometric authentication technologies such as face, finger, hand, iris and speaker recognition are commercially available today and are already coming into wide use. Recent advances in reliability and performance and cost drops make these technologies attractive solutions for many computer and network access, protection of digital content and physical access control problems.

Combination of vector quantization and hidden Markov models for Arabic speech recognition

Abstract: We present experiments performed to recognize isolated Arabic words. Our recognition system is based on a combination of the vector quantization technique at the acoustic level and Markovian modelling. Hidden Markov models (HMMs) are widely used in a number of practical applications and are especially suitable in speech recognition because of their ability to handle variability of the speech signal. In our system, a word is analysed and represented as a set of acoustic vectors, then transformed into a symbolic sequence using the vector quantizer. This observation sequence is compared to reference Markov models. The word associated with the model obtaining the highest score is declared to be the recognized word.

On-line signature verification system using probabilistic feature modelling

Abstract: Signature verification is a popular biometric authentication technique. Although the use of other biometrics like retinal patterns and fingerprints are quite prevalent, they are not simple methods. With the development of very handy digitizer tablets, signatures can be captured on-line and can be processed for verification. Since signatures of the same person can vary with time and state of mind, it is necessary to develop an efficient signature verification system. We propose a signature verification system which extracts certain dynamic features derived from velocity and acceleration of the pen together with other global parameters like total time taken and number of pen-ups. The features are modeled by fitting probability density functions, i.e., by estimating the mean and variance, which could probably take care of the variations of the features of the signatures of the same person with respect to time and state of mind.

Multi-branch and two-pass HMM modeling approaches for off-line cursive handwriting recognition

Abstract: Because of large shape variations in human handwriting, cursive handwriting recognition remains a challenging task. Usually, the recognition performance depends crucially upon the pre-processing steps, e.g. the word baseline detection and segmentation process. Hidden Markov models (HMMs) have the ability to model similarities and variations among samples of a class. In this paper, we present a multi-branch HMM modeling method and an HMM-based two-pass modeling approach. Whereas the multi-branch HMM method makes the resulting system more robust with word baseline detection, the two-pass recognition approach exploits the segmentation ability of the Viterbi algorithm and creates another HMM set and carries out a second recognition pass. The total performance is enhanced by the combination of the two recognition passes. Experiments recognizing cursive handwritten words with a 30,000-word lexicon have been carried out. The results demonstrate that our novel approaches achieve better recognition performance and reduce the relative error rate significantly.

Chart analysis and recognition in document images

Abstract: Hidden Markov models are a probabilistic modeling tool for time series data. It has been successfully applied to many areas, such as speech recognition, hand-written character recognition, etc. In this paper, we present a novel statistical approach using ergodic hidden Markov models to recognize scientific charts. We also present a newly developed feature extraction method for chart images. Unlike traditional primitive-based diagram recognition method, our approach need not recognize the graphic primitives in charts thus bypassing the recognition error problem caused by the inaccurate primitive extraction that is also a major obstacle to the construction of a general chart recognition system.

Minimum classification error training of hidden Markov models for handwriting recognition

Abstract: This paper evaluates the application of minimum classification error (MCE) training to online-handwritten text recognition based on hidden Markov models. We describe an allograph-based, character level MCE training aimed at minimizing the character error rate while enabling flexibility in writing style. Experiments on a writer-independent discrete character recognition task, covering all alpha-numerical characters and keyboard symbols, show that MCE achieves more than 30% character error rate reduction compared to the baseline maximum likelihood-based system.

Signature verification adapting to intersession variability

Abstract: In signature verification using on-line hand written data, we have cleared that the pen inclination data improves the verification rate[1]. However, the influences of intersession variability on the verification rate have not been considered yet. In order to avoid making error caused by a mismatch between his/her signature and his/her references, we applied some renewal methods to make the reference. From experimental results using signatures over 9 weeks, a reference renewal using authenticated data is effective to maintain high verification rate with 98.5%.

Robustness to expression variations in fractal-based face recognition

Abstract: Face recognition has developed into a major research area in pattern recognition and computer vision. Face recognition is different from classical pattern recognition problems such as character recognition. In classical pattern recognition, there are relatively few classes, and many samples per class. With many samples per class, algorithms can classify samples not previously seen by interpolating among the training samples. On the other hand, in face recognition, there are many individuals (classes), and only a few images (samples) per person, and algorithms must recognize faces by extrapolating from the training samples. In numerous applications there can be only one training sample (image) of each person (for real time applications). In this paper, we present an approach for expression-invariant face recognition based on fractal features.

Facial Expression Recognition with Pseudo-3D Hidden Markov Models

Abstract: We introduce a novel approach to gesture recognition, based on Pseudo-3D Hidden Markov Models (P3DHMMs). This technique is capable of integrating spatially and temporally derived features in an elegant way, thus enabling the recognition of different dynamic face-expressions. Pseudo-2D Hidden Markov Models have been utilized for two dimensional problems such as face recognition. P3DHMMs can be considered as an extension of the 2D case, where the so-called superstates in P3DHMM encapsulate P2DHMMs. With our new approach image sequences can efficiently and successfully be processed. Because the 'ordinary' training of P3DHMMs is time expensive and can destroy the 3D approach, an improved training is presented in this paper. The feasibility of the usage of P3DHMMs is demonstrated by a number of experiments on a person independent database, which consists of different image sequences of 4face-expressions from 6 persons.

Image analysis and pattern recognition in biometric technologies

Abstract: . Many biometric methods are closely connected with methods of patternrecognition and image analysis. The realization of a number of biometric technologiesrequires using the last achievements in this area. Some elements of technology based onsome methods of image analysis are demonstrated by the example of iris personidentification. From a position of organizing the educational process, laboratory works in thearea of biometric technologies allow stimulating students’ inquisitiveness in studying methodsand algorithms for image processing and pattern recognition. Key words: biometric technologies, biometric authentication, image processing,education 1. Introduction Biometric and biomedical informatics are the fast developing scientific direction, studying theprocesses of creation, transmission, reception, storage, processing, displaying and interpretationof information in all the channels of functional and signal systems of living objects which areknown to biological and medical science and practice. Modern natural sciences at presentsharply need in the updating of scientific picture of the world, and the essential contribution inthis process can be made by the biometric and biomedical methods.Only some more simple (statistical) forms of biometric and biomedical information have foundtheir application when person identification, and raised interest for these methods ofidentification can be caused by new possibilities of information technologies.So, exclusively new and not explored possibilities for verification of living objects can beexpected in eniology. A concept of electromagnetic is intensively investigated at present. Newresults have been obtained in fractal analysis, using which an attempt was taken to explainsome paradoxical phenomena such as morphogenetic field, distant cells communications,anomalously high sensitivity of organism to near-zero frequency perturbations, regulationprocesses critical dependence on the fractal features of noisy environment, etc. [Polo]. Perhaps,

Off-line hand-written character recognition using integrated 1D HMMs based on feature extraction filters

Abstract: The purpose of our research is to improve the recognition rate of an off-line handwritten character recognition system using HMM (hidden Markov model), so that we can use the system for practical application. Due to the insufficient recognition rate of ID HMM character recognition systems and the requirement for a huge number of learning samples to construct 2D HMM character recognition systems, HMM-based character recognition systems have not yet achieved sufficient recognition performance for practical use. In this research, we propose the character recognition method that integrates 4 simply structured 1D HMMs all of which are based on feature extraction using linear filters. The results of our evaluation experiment using the Hand-Printed Character Database (ETL6) showed that the first rank recognition rate of the test samples was 98.5% and that the cumulative recognition rate of top 3 candidates was 99.3%. Although our method is relatively easy to implement, it can work even better than 2D HMM method. These results show the proposed method is very effective.

Clustering and classification algorithm for computer intrusion detection

Abstract: As an important part of information security, computer intrusion detection is used to capture malicious activities occurring in computer network systems. Intrusion detection techniques fall into two general categories: anomaly detection and signature recognition, which complement each other. This research focuses on signature recognition techniques for intrusion detection. Intrusion detection data is very complex and has many attributes. Many existing signature recognition techniques for intrusion detection cannot handle huge amounts of complex data from computer network systems to detect intrusions in an automatic, scalable and incremental manner. This research proposes a scalable data mining algorithm, Clustering and Classification Algorithm - Supervised (CCA-S), for automatically and incrementally learning intrusion signatures that are then used to detect intrusions. An extension of CCA-S, called CCA-S Extended (CCA-SE), is used for processing data sets whose records have both numeric and nominal attributes. In training, based on distance and target class of the data points of normal and intrusive activities, these algorithms perform supervised clustering to group training data points into clusters. The produced clusters are used in classification to predict the target class of testing data points. Several post-processing techniques, including redistribution and a special hierarchical clustering method, are used to improve the robustness and prediction accuracy. The two algorithms are tested on two large data sets for intrusion detection, respectively. The prediction accuracy, scalability and robustness of the algorithms are analyzed and compared with those of other data mining techniques. The computation cost of the two algorithms is linear to the number of data records, while the prediction accuracy is comparable to other popular data mining algorithms and robust to the input order and noise of the training data points. The testing results demonstrate the promising performance of CCA-S and CCA-SE for intrusion detection.

Shape vector characterization of Vietnamese tones and application to automatic recognition

Abstract: The tone recognition for Vietnamese standard language (Hanoi dialect) is described. The wavelet method is used to extract the pitch (F0) from a speech signal corpus. Thus, one feature vector for tone recognition of Vietnamese is proposed. Hidden Markov models (HMMs) are then used to recognize the tones. Our results show that tone recognition seems independent of the vowel but presents better accuracy if one of both monotonous tones is used as the pitch reference base. Finally, a first try of a completely isolated word recognition engine, adapted for Vietnamese, is presented.

Codebooks for signature verification and handwriting recognition

Abstract: In this paper we consider and assess the concept of using codebooks of curves to characterise a persons handwriting. This is similar to the successful methods by which handwriting has been applied to speech recognition. The handwritten signatures are scanned as binary images at 200 dpi, thinned to a single pixel width and characterised as a set of curves. Matching of signatures is achieved using a curve similarity measure. Experiments on a set of 120 handwritten signatures from six writers (20 per writer), including some forgeries, indicate the technique has potential. Whilst it does not currently perform as well as state-of-the-art signature verifiers there are numerous improvements that can be made to the technique. A number of refinements are proposed for discussion and further research.

High-precision two-kernel Chinese character recognition in general document processing systems

Abstract: This paper proposes a general Chinese document recognition system with high recognition rate, including preprocessing, recognition kernel, and post-processing, especially for low quality images. In the preprocessing module, fast rotation transformation algorithm is proposed. Since characters are extracted for recognition engines, document images must be segmented into text blocks, text lines, and then character images. In the recognition module, two recognition engines are used to recognize the character images. The weights of these kernels and features are calculated from the relative stroke widths of character images. In the post-processing module, we calculate confidence values for different candidates and then select the most confident candidate as the OCR result. The experiments show the system we propose is very effective and efficient.

An a priori indicator of the discrimination power of discrete hidden Markov models

Abstract: During the development of a hidden Markov model based handwriting recognition system, the testing phase takes a non-negligible amount of computation time. This is especially true for real application where the lexicon size is large. In order to shorten the development process, we propose an indicator of the system discrimination power. This indicator is calculated during training and its final value is obtained at the end of the training phase, without more calculation. Its definition consists of a modification of the observation probability of the validation corpus by the trained system. Some experiments were carried out and the results show clearly the correlation between this indicator and recognition rates.

A novel hybrid face profile recognition system using the FERET and MUGSHOT databases

Abstract: Face recognition has established itself as an important subbranch of pattern recognition within the field of computer science. Many state-of-the-art systems have focused on the task of recognizing frontal views of people. We present an approach for recognizing profile views (90/spl deg/) with a system trained on transformed frontal views. The system combines an artificial neural network (ANN) and a classification process based on hidden Markov models (HAM). One of the main ideas of this system is to perform the recognition task without the use of any 3D-information of heads and faces. The presented system has been tested with subsets of the FERET and the MUGSHOT databases.

Object segmentation using an array of interconnected neural networks with local receptive fields

Abstract: We introduce an architecture for object segmentation/recognition that overcomes some limitations of classical neural networks by utilizing contextual information. An important characteristic of our model is that recognition is treated as a process of discovering a pattern rather than a one-time comparison between a pattern and a stored template. Our network implements some properties of human perception and during the recognition emulates the process of saccadic eye movements. We contrast our model to hidden Markov models in application to segmentation/recognition of handwriting and demonstrate a number of advantages.