Showing papers on "Signature recognition published in 1994"

Fundamentals in handwriting recognition

Abstract: 1: Introduction and overview of field.- Frontiers in handwriting recognition.- 2: Handwritten character recognition.- Historical review of theory and practice of handwritten character recognition.- Automatic recognition of handwritten characters.- Learning, representation, understanding and recognition of characters and words - an intelligent approach.- Digital transforms in handwriting recognition.- Pattern recognition with optimal margin classifiers.- 3: Handwritten word recognition.- On the robustness of recognition of degraded line images.- Invariant handwriting features useful in cursive script recognition.- Off-line recognition of bad quality handwritten words using prototypes.- Handwriting recognition by statistical methods.- Towards a visual recognition of cursive script.- A hierarchical handwritten word segmentation.- 4: Contextual methods in handwriting recognition.- Cursive words recognition: methods and strategies.- Hidden Markov models in handwriting recognition.- Language-level syntactic and semantic constraints applied to visual word recognition.- Verification of handwritten British postcodes using address features.- Improvement of OCR by language model.- An approximate string matching method for handwriting recognition post-processing using a dictionary.- 5: Neural networks in handwriting recognition.- Neural-net computing for machine recognition of handwritten English language text.- Cooperation of feedforward neural networks for handwritten digit recognition.- Normalisation and preprocessing for a recurrent network off-line handwriting recognition system.- 6: Architectures for handwriting.- Architectures for handwriting recognition.- 7: Databases for handwriting recognition.- Large database organization for document images.- 8: Signature recognition and verification.- A model-based dynamic signature verification system.- Algorithms for signature verification.- Handwritten signature verification: a global approach.- 9: Application of handwriting recognition.- Total approach for practical character recognition system development.- A pen-based music editor.

Hidden Markov Models in Handwriting Recognition

Abstract: Hidden Markov Models (HMM) have now became the prevalent paradigm in automatic speech recognition. Only recently, several researchers in off-line handwriting recognition have tried to transpose the HMM technology to their field after realizing that word images could be assimilated to sequences of observations. HMM’s form a family of tools for modelling sequential processes in a statistical and generative manner. Their reputation is due to the results attained in speech recognition which derive mostly from the existence of automatic training techniques and the advantages of the probabilistic framework. This article first reviews the basic concepts of HMM’s. The second part is devoted to illustrative applications in the field of off- line handwriting recognition. We describe four different applications of HMM’s in various contexts and review some of the other approaches.

Biometric recognition: our hands, eyes, and faces give us away

Abstract: The author considers how there are a growing number of applications of automated fingerprint identification. He discusses several other biometric methods of identification, including hand, facial, and eye recognition. For some applications, these methods are better than fingerprint identification, since they require smaller data signatures, may cost less, and avoid the criminal stigma of fingerprinting. >

Off-line recognition of signatures using revolving active deformable models

Abstract: This paper presents a novel approach to the problem of signature recognition. We introduce the use of revolving active deformable models as a powerful way of capturing the unique characteristics of a signature's silhouette. Experimental evidence shows that the silhouette of a signature uniquely determines the signature in the majority of cases. The objective of our method is to recognize signatures based on the spatial properties of the signature boundaries. Our active deformable models originate from the snakes introduced to computer vision by Kass et al. (1987), but their implementation has been tailored to the task at hand. These computer-generated models interact with the virtual gravity field created by the image gradient. Ideally, the uniqueness of this interaction mirrors the uniqueness of the signature's silhouette. The proposed method obviates the use of a computationally expensive segmentation approach and yields satisfactory results regarding performance, without compromising the accuracy rate. Interestingly, the active deformable models have been implemented in such a way, that the method is potentially fully parallelizable. The experiments performed with a signature database show that the proposed method is promising. >

A new composite feature vector for Arabic handwritten signature recognition

Abstract: One of the main difficulties in solving complex recognition problems is to find an optimum feature vector that translates the input image to a set of numeric values to be presented to the classifier. Optimum in the sense that it classifies samples correctly, it is easy to compute and it is small in size. This step is essential to reduce the amount of data presented to the classifier. Even if we have an excellent learning classifier, the role of the feature vector should not be underestimated. We present a comparative study for a large number of features [210] previously studied in the literature as applied to the problem of recognizing Arabic handwritten signatures. Based on the statistical results of this study, a new feature vector was suggested and tested. It yielded 98.6% recognition rate for our specific application. Since signature were represented as a 2D array of binary values output from a scanner, it is our view that the proposed vector can be generalized to the problem of recognizing any limited number of 2D binary patterns. >

Cursive script recognition using semi continuous hidden Markov models in combination with simple features

Abstract: An approach is presented to use left-to-right Hidden Markov Models adopted from speech recognition for recognition of cursive script. A sequence of feature vectors is obtained from the script line deliberately doing without preprocessing to reduce writer variability. The goal is to investigate wether or not it is possible to statistically model cursive script with HMMs for recognition based on such features.

Recognizing off-line cursive handwriting

Abstract: We present a system for recognizing off-line, cursive, English text, guided in part by global characteristics (style) of the handwriting. We introduce a new method for segmenting words into letters, based on minimizing a cost function. Segmented letters are normalized with a novel algorithm that scales different parts of a letter separately removing much of the variation in the writing. We use a neural network for letter recognition and use the output of the network as posterior probabilities of letters in the word recognition process. We found that using a hidden Markov Model for word recognition is less successful than assuming an independent process for our small set of test words. In our experiments with several hundred words, written by 7 writers, 96% of the test words were correctly segmented, 52% were correctly recognized, and 70% were in the top three choices. >

Signature recognition apparatus which can be trained with a reduced amount of sample data

Abstract: A signature recognition apparatus reduces the volume of training data needed and shortens the learning period. In the apparatus, a sample generating section generates sample data. A coupling load coefficient is determined based on the sample data, thereby obviating the need for additional sample data. The apparatus also uses a fuzzy net which implements a linear function in its output layer to shorten the learning period relative to the learning period required for a net implementing a non-linear function such as a sigmoid.

Voice and Facial Image Integration for Person Recognition

Abstract: Automatic person recognition systems attempt to use physical or behavioural characteristics in order to perform the recognition task, and to this extent may be compared with humans. The face and voice of a person are two sources of characteristics that can be provide information about the identity of an individual. We describe a person recognition system using face and voice as primary sources of personal identity information. Results of preliminary investigations on person recognition, based on still face profile images and voice, show improved recognition accuracy compared to face profile recognition and voice recognition performed separately. Further, a recognition system based on visual and acoustic speech is proposed.

Off-line recognition of cursive script produced by a cooperative writer

Abstract: A method for the off-line recognition of cursive handwriting based on hidden Markov models is described. The features used in the HMMs are based on the arcs of skeleton graphs of the words to be recognized. An average correct recognition rate of over 98% on the word level has been achieved in experiments with cooperative writers using two dictionaries of 150 words each.

Real time implementation of HMM speech recognition for telecommunications applications

Abstract: This paper describes an implementation of connected word recognition using commercially available parallel processing DSPs. The recognition system uses continuous density HMMs for speaker independent recognition over a public switched network. It describes how the computationally intensive functions can be optimised for efficient real time implementation. Results of recognition accuracy are presented for a difficult task of connected digit recognition with data from a live operator environment and an isolated digit recognition task. >

A complement to variable duration hidden Markov model in handwritten word recognition

Abstract: Because of large variation involved in handwritten words, the recognition problem is very difficult. Hidden Markov models (HMM) have been widely and successfully used both in speech and handwriting recognition. Basically, there are two strategies of using HMM: model discriminant HMM (MD-HMM) and path discriminant HMM (PD-HMM). Both of them have their advantages and disadvantages, and are discussed in this paper. Chen, Kundu and Sihari (see Proc. IEEE Int. Conference on Acoust., Speech, Signal Processing, (Minneapolis, Minnesota), p.V.105-108, April 1993) have developed a handwritten word recognition system using continuous density variable duration hidden Markov model (CDVDHMM), which belongs to the PD-HMM strategy. We describe a MD-HMM approach with the statistics derived from the CDVDHMM parameters. Detailed experiments are carried out; and the results using different approaches are compared. >

The design of an intelligent signature processing system for banking environment

Abstract: Presents the design of an intelligent signature processing system for the banking environment. The system is known as AutoSIG which provides facilities to capture, display, manipulate and edit customers' signatures to help signature verification. In addition, automatic signature recognition is also supported. The goal of the system is to enable the bank personnel to work more efficiently and at the same time, to make a better judgment on signature verification. We discuss the functional and system requirements of AutoSIG. The architecture of AutoSIG is also outlined. The AutoSIG consists of a set of tools which include a capture tool, a display tool, an image editor and an automatic signature verification tool. A signature database is used to store signature images. Image compression/decompression is also performed to reduce the storage capacity required. >

The handwritten signature in pen computing

Abstract: The handwritten signature has many purposes and meanings; underlying these related purposes there are two ideas: the handwritten signature is a distinctive personal mark, because it is distinctive, its authenticity can be verified. Both these ideas can be called into question. A handwritten signature is distinctive and verifiable only if forgery is detectable and hence impracticable. It is illuminating to contrast what an electronically-captured signature might be with its paper equivalent. Needless to say, a paper signature is bulky to store and difficult to retrieve. Much of the visual verification which is currently done is cursory and depends on subjective criteria; it is totally impracticable to apply such techniques to large volumes of signatures. Finally, the data captured on paper are relatively few. By contrast, dynamic signature capture using a digitizer gives us stroke order, pen speed and acceleration as well as the image itself. This gives the opportunity to perform biometric tests to determine those characteristic aspects of an individual's performance which are unique to him. The use of such data for verification purposes leaves a forger with more and bigger hurdles to surmount. >

View-based Models of 3D Object Recognition and Class-specific Invariances

Abstract: This paper describes the main features of a view-based model of object recognition. The model tries to capture general properties to be expected in a biological architecture for object recognition. The basic module is a regularization network in which each of the hidden units is broadly tuned to a specific view of the object to be recognized.

An Off-Line Signature Recognition Using a Fuzzy Net

Abstract: This paper presents a method of off-line signature recognition using feature strokes and a fuzzy net. For getting feature strokes a pre-processing of signature data is studied. The pre-processing consists of a thinning method and a method of extracting feature strokes from an original signature and corresponding strokes from new signatures. The structure of fuzzy net is a simple perceptron and each unit in the input layer has a fuzzy membership function. From the connection weights of the trained network, it is easy to know which features of input data are important for signature recognition, i.e. the personal characteristics of the signatures are known from the fuzzy net. An experiment is done to show the feasibility of the new method.

Handwritten character recognition by extended loop neural networks

Abstract: Presents an extended loop neural network approach to handwritten character recognition. Experiments show that this method is very effective. The recognition rate by this method is higher than that by a backpropagation network. >

A hidden Markov model approach for degraded and connected character recognition

Abstract: This paper describes a method of representing a character according to shape profiles generated from vertical and horizontal scans through the lines of pixels constituting the character image. A pair of hidden Markov models is used to capture this shape profile information. Recognition is performed by combining knowledge from these two experts such that a test image may be "scored" against all character models, and a ranked recognition result can be output for further contextual processing. The success of this method in recognising degraded printed text is demonstrated here. Due to the powerful nature of the character representation, and particularly due to the method's suitability for the recognition of contextual characters, it is expected that the method can be applicable to hand-printed character recognition. >

Handwritten word recognition using statistics

Abstract: In this paper, a system for the recognition of images of handwritten cursive words is presented. Since all the features of the described system are based on symbolic representation of the contour and skeleton, they can be computed very efficiently. The hidden Markov technique, already been used successfully for speech recognition, scores noteworthy results in handwriting recognition, too. In fact, the recognition results are better the larger the number of images contained in the training set. The system has been tested exhaustively with US city names as well as names of German cities. >

Towards a Robust Physics-Based Object Recognition System

Abstract: A successful 3D object recognition system must take into account imperfections in the input data, due for example to fragmentation or sensor noise. In this paper we propose a methodology for robust 3D object recognition using uncertain image data. In particular, we present a method capable of achieving acceptable performance in the presence of both segmentation problems and sensor uncertainty, thus eliminating the need for ad hoc heuristics. The proposed method is based upon the use of probabilistic models suggested by the underlying physics processes. These models are statistically validated and tested under controlled experimentation.

An improved HMM/VQ training procedure for speaker-independent isolated word recognition

Abstract: This paper describe an improved training procedure in a HMM/VQ speech recognition system for speaker-independent speech recognition. The phoneme based Gaussian mixture models (GMM) were generated in the first step modeling using the Expectation-Maximization (EM) algorithm. These Gaussians more accurately describe the distribution characteristic of the phonemes in the speech signal space. Therefore better first step modeling is achieved and the performance of the whole recognition system is improved. The new method was used in a speaker-independent isolated digits and phoneme recognition tasks. Two English databases were used for the training and testing. Significant improvements have been achieved in comparison with the conventional HMM/VQ system. >

Pattern recognition based on information theory principles

Abstract: One of the main problems faced in the development of pattern recognition algorithms is assessment of their performance. This paper describes the development of a novel technique for the assessment of information content of 2-D patterns encountered in practical pattern recognition problems. The technique is demonstrated by itsapplication to multi-font typed character recognition. In this work we firstly developed an information model applicable to any pattern, and its elaboration to measure recognition performance, and secondly we used this model to derive parameters such as the resolution required to distinguish between the patterns. This has resulted in a powerful method for assessing the perfoimance of any pattern recognition system.Keywords: pattern recognition, information theory, character recognition, recognition information 1. INTRODUCTION Pattern Recognition is one of the fastest growing scientific areas with applications across a wide variety of disciplines. The tasks of pattern recognition are basically to remove the need for a trained operator to perform therecognition, or to enable recognition to be performed that would otherwise be impossible [1]. When examining a

Signature recognition device

Abstract: PURPOSE:To recognize a signature hand-written by a writer through an off-line input by applying fuzzy deduction processing to a detected stroke density in each scanning direction so as to convert the stroke density into fuzzy density subject to data compression. CONSTITUTION:A memory scanning section accesses addresses in the unit of picture elements of signature data stored in a memory in various directions such as horizontal direction and vertical direction to output the result to a counter section and to obtain a stroke density of the signature data in each scanning direction as shown in figure (a). An input fuzzy section is prepare for each scanning direction of the memory scanning section and each input fuzzy section has a triangle membership function as an internal function as shown, e.g. in figure (b) and fuzzy deduction is implemented by using the membership function to convert the stroke density of the signature data into the fuzzy density for each scanning direction. Thus, a data change due to a blur of a writer in the process of writing is absorbed and the signature data easily recognized are obtained.

Speech recognition by extended loop neural network

Abstract: Presents an extended loop neural network approach to speech recognition. This speech recognition approach is characterized by the following important properties due to the associative memory neural network. (1) It has the features of great adaptivity and fault tolerance to carry out recognition. (2) The recognition system can be constructed which allows for the formation of arbitrary nonlinear decision surfaces. (3) The recognition system can perform not only the recognition task but also restore the correct information from incomplete even some extent incorrect information at the same time. Experiments are also conducted and the results show that this speech recognition approach has great application potentials. >

A discriminative hidden Markov model recognizer with neural network postprocessor

Abstract: This paper is concerned with the problem of improving recognition accuracy of hidden Markov models (HMM) for sequential pattern recognition. It is argued that maximum-likelihood estimation of the HMM parameters via the forward-backward algorithm may not lead to values which maximize recognition accuracy. We introduce a hybrid method with neural network postprocessor which is aimed at minimizing the number of recognition errors. This method exploits the discrimination capability of neural network classifier while using HMM formalism to capture the dynamics of input patterns. Although it has not been proved that the presented method is a kind of maximum mutual information estimation, experimental results with online handwriting characters suggest that it leads to fewer recognition errors than can be obtained with the conventional recognition method. >

Method for normalization of symbol

Abstract: PURPOSE: To provide a method for speedily and simply executing normalization necessary for recognizing a code and a signature. CONSTITUTION: A handwritten test code (alphabetical character, word or signature, e.g.) to be written on a digitized tablet is compared with a model code and normalized to obtain conversion necessary for suiting this test code to the model code. Next this conversion is applied to this test code. Information in the form of this test code is stored in this normalization. One feature is that the model code is a line segment and the other feature is that it is one example of the code to normalize. Normalization like this can be used for application such as character recognition, text recognition or signature recognition as a preprocessing step.

Toward an automated signature recognition toolkit for mission operations

Abstract: Signature recognition is the problem of identifying an event or events from its time series The generic problem has numerous applications to science and engineering At NASA's Johnson Space Center, for example, mission control personnel, using electronic displays and strip chart recorders, monitor telemetry data from three-phase electrical buses on the Space Shuttle and maintain records of device activation and deactivation Since few electrical devices have sensors to indicate their actual status, changes of state are inferred from characteristic current and voltage fluctuations Controllers recognize these events both by examining the waveform signatures and by listening to audio channels between ground and crew Recently the authors have developed a prototype system that identifies major electrical events from the telemetry and displays them on a workstation Eventually the system will be able to identify accurately the signatures of over fifty distinct events in real time, while contending with noise, intermittent loss of signal, overlapping events, and other complications This system is just one of many possible signature recognition applications in Mission Control While much of the technology underlying these applications is the same, each application has unique data characteristics, and every control position has its own interface and performance requirements There is a need, therefore, for CASE tools that can reduce the time to implement a running signature recognition application from months to weeks or days This paper describes our work to date and our future plans