Showing papers on "Signature recognition published in 2000"

An introduction evaluating biometric systems

Abstract: On the basis of media hype alone, you might conclude that biometric passwords will soon replace their alphanumeric counterparts with versions that cannot be stolen, forgotten, lost, or given to another person. But what if the actual performance of these systems falls short of the estimates? The authors designed this article to provide sufficient information to know what questions to ask when evaluating a biometric system, and to assist in determining whether performance levels meet the requirements of an application. For example, a low-performance biometric is probably sufficient for reducing-as opposed to eliminating-fraud. Likewise, completely replacing an existing security system with a biometric-based one may require a high-performance biometric system, or the required performance may be beyond what current technology can provide. Of the biometrics that give the user some control over data acquisition, voice, face, and fingerprint systems have undergone the most study and testing-and therefore occupy the bulk of this discussion. This article also covers the tools and techniques of biometric testing.

Handwritten sentence recognition

Abstract: We present a system for reading handwritten sentences and paragraphs The system's main components are preprocessing, feature extraction and recognition In contrast to other systems, whole lines of text are the basic units for the recognizer Thus the difficult problem of segmenting a line of text into individual words can be avoided Another novel feature of the system is the incorporation of a statistical language model into the recognizer Experiments on the database described previously by the authors (1999) have shown that a recognition rate on the word level of 795% and 6005% for small (776 words) and larger (7719 words) vocabularies can be reached These figures increase to 843% and 6732% if the top ten choices are taken into account

Harmonics load signature recognition by wavelets transforms

Abstract: Power quality has become an important concern both to the utilities and their customers. End user equipment is often more sensitive to disturbances that exist both on the supplying power system and within the customer facilities. Power quality embraces problems caused by harmonics, over or under-voltages, or supply discontinuities. Harmonics are caused by all sorts of nonlinear loads. In order to fully understand the problems caused by harmonics pollution, an effective means of identifying sources of power harmonics is important. The authors used fuzzy numbers for harmonics signature recognition. In this paper, the authors have made use of new developments in wavelets so that each type of current waveform polluted with power harmonics can well be represented by a normalised energy vector consisting of five elements. Furthermore, a mixture of harmonics load can also be represented by a corresponding vector. This paper describes the mathematics and algorithms for arriving at the vectors, forming a strong foundation for real-time harmonics signature recognition, in particular useful to the re-structuring of the whole electric power industry.

Signature recognition system and method

Abstract: A method of authenticating a signature including the steps of sampling a signature and storing data representative of the signature, converting the data to high dimensions vectors, feeding the high dimension vectors to an unsupervised neural network, performing a high order principal component extraction process on the high dimensions vectors to thereby identifying clusters of high dimension points, and analyzing the clusters of high dimension points to determine, based on previously stored information, the authenticity of the signature. Also an apparatus for such authentication including a sampling device for sampling a signature and storing data representative of the signature, a converting device connected downstream of the sampling device for converting the data to high dimension vectors, an unsupervised neural network for receiving the high dimension and performing a high order principal component extraction process on the high dimensions vectors to thereby identify clusters of high dimension points, and an analyzing device connected to the unsupervised neural network for analyzing the clusters of high dimension points to determine the authenticity of the signature.

On-line signature verification based on altitude and direction of pen movement

Abstract: A signature is widely used to authorize who issued the document. However, a signature has ambiguity, and it is difficult to distinguish the authentic signature from the mimicked signature only by the bit-mapped pattern. It is expected that the altitude and the azimuth of the gripped pen under signing depends on the shape of the writer's hand and the habit of writing. We propose a new on-line writer authentication system using the pen altitude, pen azimuth, shape of signature, and writing pressure in real time. From experimental results with writing information by 24 writers, it is found that the authentication rate of 98% is obtained.

Speech recognition driven system with selectable speech models

Abstract: A speech recognition driven system provides a speech model based on a biometric signature. Initially, the speech recognition driven system receives a biometric signature from a user of the system. Based upon the received biometric signature, the system selects a speech model. The selected speech model is utilized to determine whether a voice input provided by the user corresponds to a speech selectable task that is recognized by the speech recognition driven system. When the voice input provided by the user corresponds to the speech selectable task, the system causes the speech selectable task to be performed. In one embodiment, the biometric signature is an image of the user's face. When face recognition technology is implemented, the image of the user's face is utilized to select a speech model.

Automatic handwriting gestures recognition using hidden Markov models

Abstract: Hidden Markov models have been successfully employed in speech recognition and, more recently, in sign language interpretation. They seem adequate for visual recognition of gestures. In this paper, two problems often eluded are considered. We propose to use the Bayesian information criterion in order to determine the optimal number of model states. We describe the contribution of continuous models in opposition to symbolic ones. Experiments on handwriting gestures show recognition rate between 88% and 100%.

Writer verification using altitude and direction of pen movement

Abstract: Signature is widely used to authorize who issued the document. However, signature has ambiguity, and it is difficult to distinguish the authentic signature from the mimicked signature by using bit mapped patterns only. On the other hand, altitude and direction of the gripped pen under signing depends on the shape of writer's hand and the habit of writing. In this paper, we propose a new online writer verification method which uses the pen movement in signing. From the experimental results with 24 writers, a verification rate of 100% was obtained.

System and method for automatic voice recognition using mapping

Abstract: A method and system that combines voice recognition engines and resolves differences between the results of individual voice recognition engines using a mapping function. Speaker independent voice recognition engines and speaker-dependent voice recognition engines are combined. Hidden Markov Model (HMM) engines and Dynamic Time Warping (DTW) engines are combined.

Wavelet feature vectors for neural network based harmonics load recognition

Abstract: Power quality embraces problems caused by harmonics, over or under-voltages, or supply discontinuities. Harmonics are caused by all sorts of non-linear loads. In order to fully understand the problems, an effective means of identifying sources of power harmonics is important. In this paper, the authors make use of new developments in wavelets so that each type of current waveform polluted with power harmonics can well be represented by a normalised energy vector consisting of five elements. Furthermore, a mixture of harmonics load can also be represented by a corresponding vector. This paper describes the mathematics and algorithms for arriving at the vectors, forming a strong foundation for real-time harmonics signature recognition, in particular, useful to the re-structuring of the whole electric power industry. The system performs exceptionally well with the aid of an artificial neural network.

System and method of biometric authentication of electronic signatures using iris patterns

Abstract: The use of iris recognition to authenticate the signatory to an electronic document is provided. A system and method are described which permit capture of handwritten graphic signatures and true identity through an iris-based biometric and association of these data with electronic documents. The system and method include capture and storage of a powerful biometric identifier based on the iris of the eye which uniquely identifies and binds the signatory to the signature and the document. A biometric record is produced which contains information about the document, such as, for example, the conditions under which it was signed, the reason for signing as understood by the signatory, the biometric template of the signatory, and a graphic representation of the signature. Stored with the document, this biometric record allows later detection of fraud associated with the signature, including forgery, replacement of the signature, alteration of the document, or alteration of the signature object itself.

Improved segmentation and segment association for on-line signature verification

Abstract: Segmentation is a process that partitions a signature into different segments. This enables local characteristics pertinent to the signature to be extracted and subsequently compared. There are many variations in the choice of the segmentation technique depending on the underlying verification strategy. Immunity to intra-signer variations poses the greatest challenge to most techniques employed. This paper proposes a novel segmentation technique that chooses to segment signatures at relatively non-complex shape sections. Through the process of association, segments of an instance are matched with those of another instance to yield corresponding segments for local comparisons. An important aspect of the method is the reduction in sensitivity to intra-personal variations. Experimental results are presented to demonstrate the effectiveness of this approach to the segmentation of signatures.

Diagnosing transmission line termination faults by means of wavelet based crosstalk signature recognition

Abstract: This paper describes a technique that allows one to identify the faulty condition (open or short circuit) at a termination of a multiconductor transmission line structure by measuring the induced voltage at the other end. The wavelet theory is used to filter out from the signal the components due to unwanted sources, and to decompose it to obtain the fault's signature. The comparison (or matching) algorithm is substituted by an artificial neural network. Two differently designed neural networks are used to validate the results and the overall procedure is also tested on an experimental set-up.

Off-line handwritten Chinese character recognition with hidden Markov models

Abstract: Off-line handwritten Chinese character recognition is one of the most difficult tasks of optical character recognition because of complexity of patterns, large quantity of classes, many uncertainties, etc. The hidden Markov model (HMM) method has achieved great success in the field of speech recognition. It also exhibits potential advantage in degraded text and handwritten character recognition. We present a modeling and recognition method of off-line handwritten Chinese character with hidden Markov models and its experimental result.

Improved degraded document recognition with hybrid modeling techniques and character n-grams

Abstract: A robust multifont character recognition system for degraded documents, such as photocopy or fax, is described. The system is based on hidden Markov models using discrete and hybrid modeling techniques, where the latter makes use of an information theory-based neural network. The presented recognition results refer to the SEDAL-database of English documents using no dictionary. It is also demonstrated that the usage of a language model that consists of character n-grams yields significantly better recognition results. Our resulting system clearly outperforms commercial systems and leads to further error rate reductions compared to previous results reached on this database.

Apparatus and method for utilizing multiple digital image representations of a document

Abstract: A document processing system includes an image capture system, a multiple engine recognition system, and an application system. The image capture system generates a first electronic image of a first region of a document, where the first electronic image has a first image format. The image capture system also generates a second electronic image of a second region of a document, where the second electronic image has a second image format. The multiple recognition system transmits the first electronic image to a first recognition engine and a second recognition engine. The multiple engine recognition system also transmits the second electronic image to the second recognition engine. The first recognition engine generates a first recognition result, and the second recognition engine generates a second recognition result. The recognition system further combines the first recognition result and the second recognition result into a final recognition result. The application system transmits the first electronic image and the second electronic image from the image capture system to the recognition system and retrieves the final recognition result from the recognition system. Utilizing multiple recognition engines and multiple image formats allows greater customization of the final recognition result, and therefore improves recognition rates.

Online handwritten formula recognition with integrated correction recognition and execution

Abstract: Presents the extension of an approach for online handwritten formula recognition. The introduction of some constraints concerning the handwriting production process and the robust hidden Markov model (HMM) framework yields recognition rates up to 97.7%. The high, but still limited recognition rate demonstrates the user's need for some correction facilities, in order to modify misclassified symbols and thus to avoid the re-writing of the entire expression. Such facilities can further be used to provide the opportunity to develop an expression on the electronic paper, as it is often desired from a practical point of view.

Free-form 3D object compression using surface signature

Abstract: This paper introduces a new transform, called the signature transform, to concisely represent fee-from 3D objects. The signature transform is based on a free-form surface representation called the surface signature. The surface signature captures some information about a 3D surface, as viewed from a special point called the anchor, such as the curvature, distance from anchor point,....etc. The surface signature stores this information in the form of a 2D image called the surface signature image. The signature transform uses different variations of the surface signature as viewed from selected landmark points. The selection of anchor points is crucial to the success of the signature transform an approach for selecting landmark points based on curvature value will be presented. The signature transform can then be used as a form of a progressive compression of objects that will allow the view and manipulation of the 3D object even if all the compression data are not received. Unlike the previously existing progressive compression techniques, the signature transform does not require receiving the data in special order nor does it have key frames in the representation.© (2000) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Generation of optimum signature base sequences for speech signals

Abstract: In previous publications, we proposed a novel method to represent signals in terms of, so called, "signature base functions-SBF" which were extracted from the physical features of the waveform under consideration. SBF were determined in ad-hoc manner, which requires tedious search process, and they were not orthogonal. Furthermore, optimality of SBF was in question. In this work however, we suggest a well-organised procedure to generate "optimum orthogonal signature base functions-OSBF" for selected waveforms, which in turn provides excellent means for signal representations. It is shown that the new method of signal representation, which is based on OSBF, requires less computation time with substantial signal compression and results in efficient speaker dependent recognition.

Human facial expression recognition based on learning subspace method

Abstract: The learning subspace method (LSM) is one of the most important methods for pattern recognition and it has been successfully used in many practical applications. We propose to use the LSM for human facial expression recognition. Seven expression subspaces are built for expression models. The idea of recognizing facial expression through a single static image is realized and the recognition rate as high as 89.5% is achieved. In order to make these expression subspaces more adaptive we can gradually learn them by using the averaged learning subspace method (ALSM). Experimental results also indicate that the recognition rate is over 90%. The dynamic characteristics of the projection vector sequence on these facial expression subspaces are also discussed.

Stochastic target detection for hyperspectral data

Abstract: There has been considerable interest in the recognition and identification of known materials and objects by using airborne hyperspectral sensors. Hyperspectral sensors provide the spectral signature for every pixel, which can be compared to the signature of a material of interest. In this paper a signature recognition algorithm is developed based on the generalized likelihood ratio test (GLRT) approach. Our starting model for target and clutter assumes that the target signature replaces the background and does not add to it. The recognition algorithm is developed using this model, and then applied to hyperspectral data to illustrate the performance.

Device and method for biometric authentication

Abstract: FIELD: biometric authentication. SUBSTANCE: method includes saving several sets of standard data, received on basis of biometric data related to biometric parameter, biometric data are registered, registered data are converted with use of certain algorithm into control data and for authentication these are compared between to control data. Several sets of standard data are received by using different algorithms, at least two. As biometric parameters, parameters of retina, face, speech, fingerprints or signature can be used, as well as hand motion dynamics during writing of a signature. EFFECT: higher efficiency. 4 cl, 2 dwg

An on-line Chinese signature verification system using an impulse response of signature generation model

Abstract: Chinese signature verification is the leading edge in the field of Chinese information processing. This paper presents a function method for signature verification using an impulse response of signature generation model, which is more strict than the parameter method that is used widely. In order to avoid high order inverse matrix calculating, an approach of segmentation proceed with the administrative levels of Chinese character frames, and stroke-based segmentation is used. So that an impulse response of signature generation model can be used for signature verification. The experiments reveal the feasibility of the method.

Signature recognition system and method

Abstract: (57) Abstract: A method for authenticating a signature, by sampling the signature, and storing data representative of a signature, and converting the data into high-dimensional vector, unmanaged neural network of high-dimensional vector and sending to, perform the higher order principal component extraction process in the high-dimensional vector, thereby identifying a cluster of points in higher dimensions, to analyze clusters of points higher dimensions, information previously stored and determining the authenticity of the signature based on a process comprising. Also a device for such authentication, by sampling the signature, and a sampling device for storing data representing the signature, is connected downstream of the sampling device to convert the data into high-dimensional vector and conversion apparatus, and receiving a high-dimensional, running order principal component extraction process in the high-dimensional vector, thereby, the unmanaged neural network for identifying the cluster of points higher dimensions, in terms of high-dimensional analyzing the cluster, for determining the authenticity of the signature, and analyzing means connected to the unmanaged neural network, a device comprising a.

Research on nonspecific vocabulary speech recognition

Abstract: In automatic speech recognition, speech problems are related to nonspecific vocabulary, i.e. nonspecific vocabulary speech recognition. In this paper, some research on the algorithm for nonspecific vocabulary speech recognition has been done. In order to get excellent recognition result, the breadth-first dynamic time-warping algorithm is set forth and implemented. The algorithm gives more considerations on breadth of search, therefore, it avoids being trapped in local minimum point. Because no specific vocabulary is given, it is impossible to apply statistic method in pattern recognition, and the model match method is the only choice. Due to the consideration about recognition ratio, the method of fuzzy pattern recognition has been adopted The classification of different pattern is based on degrees of membership of these patterns, and the degrees of membership is calculated according to the accumulated distance between the recognized order of speech vector and the order of speech vector in speech model base.