Showing papers on "Handwriting recognition published in 2002"

The IAM-database: an English sentence database for offline handwriting recognition

Abstract: In this paper we describe a database that consists of handwritten English sentences. It is based on the Lancaster-Oslo/Bergen (LOB) corpus. This corpus is a collection of texts that comprise about one million word instances. The database includes 1,066 forms produced by approximately 400 different writers. A total of 82,227 word instances out of a vocabulary of 10,841 words occur in the collection. The database consists of full English sentences. It can serve as a basis for a variety of handwriting recognition tasks. However, it is expected that the database would be particularly useful for recognition tasks where linguistic knowledge beyond the lexicon level is used, because this knowledge can be automatically derived from the underlying corpus. The database also includes a few image-processing procedures for extracting the handwritten text from the forms and the segmentation of the text into lines and words.

An overview and comparison of voting methods for pattern recognition

Abstract: In pattern recognition, there is a growing use of multiple classifier combinations with the goal to increase recognition performance. In many cases, plurality voting is a part of the combination process. In this article, we discuss and test several well known voting methods from politics and economics on classifier combination in order to see if an alternative to the simple plurality vote exists. We found that, assuming a number of prerequisites, better methods are available, that are comparatively simple and fast.

Method of combining data entry of handwritten symbols with displayed character data

Abstract: A pen or stylus-operated graphical user interface for a computer or computing device, which includes a sensing surface having an area corresponding to a data input field, the data input field being conditioned for hand entering and editing of graphical input symbols, and handwriting recognition software operative to analyze the graphical input symbols and superimposing a display field of character data corresponding to the graphical input symbols on the data input field.

Integration of speech and stylus input to provide an efficient natural input experience

Abstract: A user interface that accepts input data through both speech and the use of a pen or stylus. With the interface, a user can employ voice recognition to enter a large volume of data, and subsequently employ a stylus input to modify the input data. A user can also employ stylus input, such as data from a handwriting or character recognition operation, to control how subsequently spoken words are recognized by a voice recognition operation. Further, a user may input data using a stylus, and then modify the input data using a voice recognition operation. A user may also employ a voice recognition operation to control how handwriting or character data input through a stylus is recognized by a handwriting recognition operation or a character recognition operation. In addition to a user interface, a technique is disclosed for inputting data into a computer where information is shared between a speech input operation and a handwriting input operation.

A data base for Arabic handwritten text recognition research

Abstract: In this paper we present a new database for off-line Arabic handwriting recognition, together with associated preprocessing procedures. We have developed a new database for the collection, storage and retrieval of Arabic handwritten text (AHDB). This is an advance both in terms of the size of the database as well as the number of different writers involved. We further designed an innovative, simple yet powerful, in place tagging procedure for our database. It enables us to easily extract the bitmaps of words. We also constructed a preprocessing class, which contains some useful preprocessing operations. In this paper the most popular words in Arabic writing were identified for the first time, using an associated program.

Feature selection using multi-objective genetic algorithms for handwritten digit recognition

Abstract: Discusses the use of genetic algorithms for feature selection for handwriting recognition. Its novelty lies in the use of multi-objective genetic algorithms where sensitivity analysis and neural networks are employed to allow the use of a representative database to evaluate fitness and the use of a validation database to identify the subsets of selected features that provide a good generalization. Comprehensive experiments on the NIST database confirm the effectiveness of the proposed strategy.

Automatic segmentation of the IAM off-line database for handwritten English text

Abstract: Presents an automatic segmentation scheme for cursive handwritten text lines using the transcriptions of the text lines and a hidden Markov model (HMM) based recognition system. The segmentation scheme has been developed and tested on the IAM database that contains offline images of cursively handwritten English text. The original version of this database contains ground truth for complete lines of text only, but not for individual words. With the method described in the paper the usability of the database is greatly improved because accurate bounding box information and ground truth for individual words (including punctuation characters) is now available as well. Applying the segmentation scheme on 417 pages of handwritten text a correct word segmentation rate of 98% has been achieved, producing correct bounding boxes for over 25,000 handwritten words.

Baseline estimation for Arabic handwritten words

Abstract: Baseline information has been used for diverse purposes in handwriting research. The baseline represents a first orientation in a word and it is often a precondition for subsequent algorithms, including preprocessing tasks, segmentation and feature extraction for recognition systems. Approaches based on the horizontal projection histogram are used for Arabic printed text but they are ill-suited for Arabic handwritten words. In this paper we present a method that is completely based on polygonally approximated skeleton processing. The central algorithm is concerned with finding features in the skeleton and processing linear regression analysis. Our method performs very well as long as the model assumption of one straight line applies. We tested the method on 26459 isolated Tunisian town names written by 411 writers (IFNIENIT-database).

Writer adaptation for online handwriting recognition

Abstract: Writer-adaptation is the process of converting a writer-independent handwriting recognition system into a writer-dependent system. It can greatly increasing recognition accuracy, given adequate writer models. The limited amount of data a writer provides during training constrains the models' complexity. We show how appropriate use of writer-independent models is important for the adaptation. Our approach uses writer-independent writing style models (lexemes) to identify the styles present in a particular writer's training data. These models are then updated using the writer's data. Lexemes in the writer's data for which an inadequate number of training examples is available are replaced with the writer-independent models. We demonstrate the feasibility of this approach on both isolated handwritten character recognition and unconstrained word recognition tasks. Our results show an average reduction in error rate of 16.3 percent for lowercase characters as compared against representing each of the writer's character classes with a single model. In addition, an average error rate reduction of 9.2 percent is shown on handwritten words using only a small amount of data for adaptation.

Chinese character handwriting recognition system

Abstract: A handwritten Chinese character input method and system is provided to allow users to enter Chinese characters to a data processor by adding less than three strokes and one selection movement such as mouse clicking or stylus or finger tapping. The system is interactive, predictive, and intuitive to use. By adding one or two strokes which are used to start writing a Chinese character, or in some case even no strokes are needed, users can find a desired character from a list of characters. The list is context sensitive. It varies depending on the prior character entered. Compared to other existing systems, this system can save users considerable time and efforts to entering handwritten characters.

Feature extraction with wavelet transform for recognition of isolated handwritten Farsi/Arabic characters and numerals

Abstract: A system is developed for recognition of handwritten Farsi/Arabic characters and numerals. The discrete wavelet transform is utilized to produce wavelet coefficients, which are used for classification. We used Haar wavelet for feature extraction in this system. The extracted features are used as training inputs to a feed forward neural network using the backpropagation learning rule. The learning and test patterns were gathered from various people with different educational backgrounds and different ages. We categorize 32 characters in Farsi language to 8 different classes in which characters of each class are very similar to each others. There are ten digits in Farsi/Arabic languages, but two of them are not used in postal codes in Iran, so we have 8 different extra classes for digits. This system yields the classification rates of 92.33% and 91.81% for these 8 classes of handwritten Farsi characters and numerals respectively. We used this system for recognizing the handwritten postal addresses which contain the names of cities and their postal codes. Our database contains 579 postal addresses in Iran. The system yields a recognition rate of 97.24% for these postal addresses.

Hidden Markov model length optimization for handwriting recognition systems

Abstract: This paper investigates the use of three different schemes to optimize the number of states of linear left-to-right hidden Markov models (HMM). In the first method, we describe the fixed length modeling scheme where each character model is assigned the same number of states. The second method considered is the Bakis length modeling where the number of model states is set to a given fraction of the average number of observations of the corresponding character. In the third modeling scheme the number of model states is set to a specified quantile of the corresponding character length histogram. This method is called quantile length modeling. A comparison of different length modeling schemes was carried out with a handwriting recognition system using off-line images of cursively handwritten English words from the IAM database. For the fixed length modeling, a recognition rate of 61% was achieved. Using the Bakis or quantile length modeling the word recognition rates could be improved to over 69%.

Learning-based cursive handwriting synthesis

Abstract: In this paper an integrated approach for modeling, learning and synthesizing personal cursive handwriting is proposed. Cursive handwriting is modeled by a tri-unit handwriting model, which focuses on both the handwritten letters and the interconnection strokes of adjacent letters. Handwriting strokes are formed from generative models that are based on control points and B-spline curves. In the two-step learning process, a template-based matching algorithm and a data congealing algorithm are first proposed to extract training vectors from handwriting samples, and then letter style models and concatenation style models are trained separately. In the synthesis process, isolated letters and ligature strokes are generated from the learned models and concatenated with each other to produce the whole word trajectory, with guidance from a deformable model. Experimental results show that the proposed system can effectively learn the individual style of cursive handwriting and has the ability to generate novel handwriting of the same style.

An unconstrained handwriting recognition system

Abstract: In this paper, an integrated offline recognition system for unconstrained handwriting is presented. The proposed system consists of seven main modules: skew angle estimation and correction, printed-handwritten text discrimination, line segmentation, slant removing, word segmentation, and character segmentation and recognition, stemming from the implementation of already existing algorithms as well as novel algorithms. This system has been tested on the NIST, IAM-DB, and GRUHD databases and has achieved accuracy that varies from 65.6% to 100% depending on the database and the experiment.

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.

Combination of local and global vision modelling for Arabic handwritten words recognition

Abstract: We propose an Arabic handwritten word recognition system based on the idea of the PERCEPTRO system developed by Cote (Cote et al. (1998)) for Latin word recognition. It is a specific neural network, named transparent neural network, combining a global and a local vision modeling (GVM-LVM) of the word. In the forward propagation movement, the former (GVM) proposes a list of structural features characterizing the presence of some letters in the word. GVM proposes a list of possible letters and words containing these characteristics. Then, in the backpropagation movement, these letters are confirmed or not according to their proximity with corresponding printed letters. The correspondence between the letter shapes and the corresponding printed letters is performed by LVM using the correspondence of their Fourier descriptors, playing the role of a letter shape normalizer.

A statistical approach for phrase location and recognition within a text line: an application to street name recognition

Abstract: We describe an approach to conjointly locate and recognize a street name within a street line. The system developed is based on a probabilistic framework that naturally integrates various knowledge sources to emit a final decision. At the handwriting signal level, hidden Markov models are extensively used to provide the needed matching scores. Several optimization techniques are employed to speed up the processing time. Experiments carried out on large data sets of street line images, automatically extracted from real French mail envelope images, show very promising results.

Apparatus and methods for hand motion tracking and handwriting recognition

Abstract: Apparatus and method for handwriting recognition including a hand-held device including at least one accelerometer providing an acceleration indicating output, computation circuitry receiving the acceleration indicating output and providing a velocity indicating output and a handwritten multiple character recognizer receiving the velocity indicating output and providing a multiple character recognition output indication.

An Evaluation of Text-Entry in Palm OS – Graffiti and the Virtual Keyboard

Abstract: The handheld organizer or personal digital assistant (PDA) is rapidly becoming a popular organizational tool, and there is a need for evaluation of alphanumeric character entry on these devices. The Palm operating system, the most common PDA operating system on the market, uses two methods for character entry, an on-screen virtual keyboard and a single-character handwriting recognition system called Graffiti. An initial experiment was conducted to investigate the character entry rates of novice and expert users of the device for the two methods of input. Experts were found to reach an average rate of 21 words per minute (wpm) using Graffiti and 18 wpm using the virtual keyboard. Novices were able to use Graffiti at a rate of 7 wpm and the virtual keyboard at 16 wpm. These character entry rates are evaluated with respect to some theoretical limitations, a predicted rate of entry based on Fitts' and the Hick-Hyman laws for the virtual keyboard, and pen and paper printing for Graffiti. The potential gain for...

On-line recognition of handwritten Arabic characters using a Kohonen neural network

Abstract: Neural networks have been applied to various pattern classification and recognition problems for their learning ability, discrimination power and generalization ability The neural network most referenced in the pattern recognition literature are the multi-layer perceptron, the Kohonen associative memory and the Capenter-Grossberg ART network. The Kohonen memory runs an unsupervised clustering algorithm. It is easily trained and has attractive properties such as topological ordering and good generalization. In this study an on-line system for the recognition of handwriting Arabic characters using a Kohonen network is investigated. The input of the neural network is a feature vector of elliptic Fourier coefficients extracted from the handwritten dynamic representation. Experimental results show that the network successfully recognizes both clearly and roughly written characters with good performance.

Visual input for pen-based computers

Abstract: The design and implementation of a camera-based, human-computer interface for acquisition of handwriting is presented. The camera focuses on a standard sheet of paper and images a common pen; the trajectory of the tip of the pen is tracked and the contact with the paper is detected. The recovered trajectory is shown to have sufficient spatio-temporal resolution and accuracy to enable handwritten character recognition. More than 100 subjects have used the system and have provided a large and heterogeneous set of examples showing that the system is both convenient and accurate.

A hybrid large vocabulary handwritten word recognition system using neural networks with hidden Markov models

Abstract: We present a hybrid recognition system that integrates hidden Markov models (HMM) with neural networks (NN) in a probabilistic framework. The input data is processed first by a lexicon-driven word recognizer based on HMMs to generate a list of the candidate N-best-scoring word hypotheses as well as the segmentation of such word hypotheses into characters. An NN classifier is used to generate a score for each segmented character and in the end, the scores from the HMM and the NN classifiers are combined to optimize performance. Experimental results show that for an 80,000-word vocabulary, the hybrid HMM/NN system improves by about 10% the word recognition rate over the HMM system alone.

Combination of multiple classifiers for handwritten word recognition

Abstract: Due to large shape variations in human handwriting, recognition accuracy of cursive handwritten word is hardly satisfying using a single classifier. In this paper we introduce a framework to combine results of multiple classifiers and present an intuitive run-time weighted opinion pool combination approach for recognizing cursive handwritten words with a large size vocabulary. The individual classifiers are evaluated run-time dynamically. The final combination is weighted according to their local performance. For an open vocabulary recognition task, we use the ROVER algorithm to combine the different strings of characters provided by each classifier. Experimental results for recognizing cursive handwritten words demonstrate that our new approach achieves better recognition performance and reduces the relative error rate significantly.

Signature verification using ART-2 neural network

Abstract: The ART neural network models have been developed for the clustering of input vectors and have been commonly used as unsupervised learned classifiers. We describe the use of the ART-2 neural network model for signature verification. The biometric data of all signatures were acquired by a special digital data acquisition pen and fast wavelet transformation was used for feature extraction. The part of authentic signature data was used for training the ART verifier. The architecture of the verifier and achieved results are discussed and ideas for future research are also suggested.

Integration of an on-line handwriting recognition system in a smart phone device

Abstract: We present the evolution of our academic development to a technology driven application: the integration of unconstraint cursive on-line handwritten characters into a smart phone device. The ultimate goal of this work is to implement an accurate handwriting recognizer into mobile devices with limited computing and memory resources. Hierarchical fuzzy modeling is used to obtain a compact and robust knowledge representation and the decision process is based on an adapted fuzzy inference system to reduce computing without decreasing the performance. We describe the basic architecture of the recognition system called "ResifCar", its practical adaptation to the mobile device constraints and the recognition rates both on cursive isolated letters (91.9%) and on isolated digits (92.3%) in a writer independent context based on 100 different writers.

Large vocabulary off-line handwritten word recognition

Abstract: Considerable progress has been made in handwriting recognition technology over the last few years. Thus far, handwriting recognition systems have been limited to small-scale and very constrained applications where the number of different words that a system can recognize is the key point for its performance. The capability of dealing with large vocabularies, however, opens up many more applications. In order to translate the gains made by research into large and very-large vocabulary handwriting recognition, it is necessary to further improve the computational efficiency and the accuracy of the current recognition strategies and algorithms. In this thesis we focus on efficient and accurate large vocabulary handwriting recognition. The main challenge is to speedup the recognition process and to improve the recognition accuracy. However, these two aspects are in mutual conflict. It is relatively easy to improve recognition speed while trading away some accuracy. But it is much harder to improve the recognition speed while preserving the accuracy. First, several strategies have been investigated for improving the performance of a baseline recognition system in terms of recognition speed to deal with large and very-large vocabularies. Next, we improve the performance in terms of recognition accuracy while preserving all the original characteristics of the baseline recognition system: omniwriter, unconstrained handwriting, and dynamic lexicons. The main contributions of this thesis are novel search strategies and a novel verification approach that allow us to achieve a 120 speedup and 10% accuracy improvement over a state-of-art baseline recognition system for a very-large vocabulary recognition task (80,000 words). The improvements in speed are obtained by the following techniques: lexical tree search, standard and constrained lexicon-driven level building algorithms, fast two-level decoding algorithm, and a distributed recognition scheme. The recognition accuracy is improved by post-processing the list of the candidate N-best-scoring word hypotheses generated by the baseline recognition system. The list also contains the segmentation of such word hypotheses into characters. A verification module based on a neural network classifier is used to generate a score for each segmented character and in the end, the scores from the baseline recognition system and the verification module are combined to optimize performance. A rejection mechanism is introduced over the combination of the baseline recognition system with the verification module to improve significantly the word recognition rate to about 95% while rejecting 30% of the word hypotheses.

Digital ink database searching using handwriting feature synthesis

Abstract: A system and method for ink database searching using handwriting feature synthesis is disclosed which allows a digital ink database to be searched using a text-based query. Using a writer-specific handwriting model derived from a handwriting recognition system or suitable training procedure, a text query is converted into feature vectors that are similar to the feature vectors that would have been extracted had the author of the digital ink database written the text query by hand. The feature vectors are then used to search the database. This allows the searching of a digital ink database when the only input mechanism available is text entry, and can allow a person other than the author of the digital ink database to search the digital ink database.

Automatic detection of handwriting forgery

Abstract: We investigated the detection of handwriting forgery by both human and machine. We obtained experimental handwriting data from subjects writing samples in their natural style and writing forgeries of other subjects' handwriting. These handwriting samples were digitally scanned and stored in an image database. We investigated the ease of forging handwriting, and found that many subjects can successfully forge the handwriting of others in terms of shape and size by tracing the authentic handwriting. Our hypothesis is that the authentic handwriting samples provided by subjects in their own natural writing style will have smooth ink traces, while forged handwritings will have wrinkly traces. We believe the reason for this is that forged handwriting is often either traced or copied slowly and is therefore more likely to appear wrinkly when scanned with a high-resolution scanner. Using seven handwriting distance features, we trained an artificial neural network to achieved 89% accuracy on test samples.