Confidence- and margin-based MMI/MPE discriminative training for off-line handwriting recognition

We present a novel confidence- and margin-based discriminative training approach for model adaptation of a hidden Markov model (HMM)-based handwriting recognition system to handle different handwriting styles and their variations. Most current approaches are maximum-likelihood (ML) trained HMM systems and try to adapt their models to different writing styles using writer adaptive training, unsupervised clustering, or additional writer-specific data. Here, discriminative training based on the maximum mutual information (MMI) and minimum phone error (MPE) criteria are used to train writer-independent handwriting models. For model adaptation during decoding, an unsupervised confidence-based discriminative training on a word and frame level within a two-pass decoding process is proposed. The proposed methods are evaluated for closed-vocabulary isolated handwritten word recognition on the IFN/ENIT Arabic handwriting database, where the word error rate is decreased by 33% relative compared to a ML trained baseline system. On the large-vocabulary line recognition task of the IAM English handwriting database, the word error rate is decreased by 25% relative.     

Minimum classification error training for online handwriting recognition

This paper describes an application of the minimum classification error (MCE) criterion to the problem of recognizing online unconstrained-style characters and words. We describe an HMM-based, character and word-level MCE training aimed at minimizing the character or word error rate while enabling flexibility in writing style through the use of multiple allographs per character. Experiments on a writer-independent character recognition task covering alpha-numerical characters and keyboard symbols show that the MCE criterion achieves more than 30 percent character error rate reduction compared to the baseline maximum likelihood-based system. Word recognition results, on vocabularies of 5k to 10k, show that MCE training achieves around 17 percent word error rate reduction when compared to the baseline maximum likelihood system.     

Offline recognition of unconstrained handwritten texts using HMMs and statistical language models

This paper presents a system for the offline recognition of large vocabulary unconstrained handwritten texts. The only assumption made about the data is that it is written in English. This allows the application of Statistical Language Models in order to improve the performance of our system. Several experiments have been performed using both single and multiple writer data. Lexica of variable size (from 10,000 to 50,000 words) have been used. The use of language models is shown to improve the accuracy of the system (when the lexicon contains 50,000 words, the error rate is reduced by approximately 50 percent for single writer data and by approximately 25 percent for multiple writer data). Our approach is described in detail and compared with other methods presented in the literature to deal with the same problem. An experimental setup to correctly deal with unconstrained text recognition is proposed.     

Off-line handwritten Chinese character recognition as a compoundBayes decision problem

A handwritten Chinese character off-line recognizer based on contextual vector quantization (CVQ) of every pixel of an unknown character image has been constructed. Each template character is represented by a codebook. When an unknown image is matched against a template character, each pixel of the image is quantized according to the associated codebook by considering not just the feature vector observed at each pixel, but those observed at its neighbors and their quantization as well. Structural information such as stroke counts observed at each pixel are captured to form a cellular feature vector. Supporting a vocabulary of 4616 simplified Chinese characters and alphanumeric and punctuation symbols, the writer-independent recognizer has an average recognition rate of 77.2 percent. Three statistical language models for postprocessing have been studied for their effectiveness in upgrading the recognition rate of the system. Among them, the CVQ-based language model is the most effective one upgrading the recognition rate by 10.4 percent on the average     

Writer adaptation via deeply learned features for online Chinese handwriting recognition

This paper proposes a novel framework of writer adaptation based on deeply learned features for online handwritten Chinese character recognition. Our motivation is to further boost the state-of-the-art deep learning-based recognizer by using writer adaptation techniques. First, to perform an effective and flexible writer adaptation, we propose a tandem architecture design for the feature extraction and classification. Specifically, a deep neural network (DNN) or convolutional neural network (CNN) is adopted to extract the deeply learned features which are used to build a discriminatively trained prototype-based classifier initialized by Linde–Buzo–Gray clustering techniques. In this way, the feature extractor can fully utilize the useful information of a DNN or CNN. Meanwhile, the prototype-based classifier could be designed more compact and efficient as a practical solution. Second, the writer adaption is performed via a linear transformation of the deeply learned features which is optimized with a sample separation margin-based minimum classification error criterion. Furthermore, we improve the generalization capability of the previously proposed discriminative linear regression approach for writer adaptation by using the linear interpolation of two transformations and adaptation data perturbation. The experiments on the tasks of both the CASIA-OLHWDB benchmark and an in-house corpus with a vocabulary of 20,936 characters demonstrate the effectiveness of our proposed approach.     

Deformation models for image recognition

We present the application of different nonlinear image deformation models to the task of image recognition. The deformation models are especially suited for local changes as they often occur in the presence of image object variability. We show that, among the discussed models, there is one approach that combines simplicity of implementation, low-computational complexity, and highly competitive performance across various real-world image recognition tasks. We show experimentally that the model performs very well for four different handwritten digit recognition tasks and for the classification of medical images, thus showing high generalization capacity. In particular, an error rate of 0.54 percent on the MNIST benchmark is achieved, as well as the lowest reported error rate, specifically 12.6 percent, in the 2005 international ImageCLEF evaluation of medical image specifically categorization.     

Unsupervised writer adaptation applied to handwritten text recognition

This paper deals with the problem of off-line handwritten text recognition. It presents a system of text recognition that exploits an original principle of adaptation to the handwriting to be recognized. The adaptation principle is based on the automatic learning, during the recognition, of the graphical characteristics of the handwriting. This on-line adaptation of the recognition system relies on the iteration of two steps: a word recognition step that allows to label the writer's representations (allographs) on the whole text and a re-evaluation step of character models. Tests carried out on a sample of 15 writers, all unknown by the system, show the interest of the proposed adaptation scheme since we obtain during iterations an improvement of recognition rates both at the letter and the word levels.     

A multiple agent architecture for handwritten text recognition

This paper investigates the automatic reading of unconstrained omni-writer handwritten texts. It shows how to endow the reading system with learning faculties necessary to adapt the recognition to each writer's handwriting. In the first part of this paper, we explain how the recognition system can be adapted to a current handwriting by exploiting the graphical context defined by the writer's invariants. This adaptation is guaranteed by activating interaction links over the whole text between the recognition procedures of word entities and those of letter entities. In the second part, we justify the need of an open multiple-agent architecture to support the implementation of such a principle of adaptation. The proposed platform allows to plug expert treatments dedicated to handwriting analysis. We show that this platform helps to implement specific collaboration or cooperation schemes between agents which bring out new trends in the automatic reading of handwritten texts.     

Extraction and optimization of B-spline PBD templates forrecognition of connected handwritten digit strings

The recognition of connected handwritten digit strings is a challenging task due mainly to two problems: poor character segmentation and unreliable isolated character recognition. The authors first present a rational B-spline representation of digit templates based on Pixel-to-Boundary Distance (PBD) maps. We then present a neural network approach to extract B-spline PBD templates and an evolutionary algorithm to optimize these templates. In total, 1000 templates (100 templates for each of 10 classes) were extracted from and optimized on 10426 training samples from the NIST Special Database 3. By using these templates, a nearest neighbor classifier can successfully reject 90.7 percent of nondigit patterns while achieving a 96.4 percent correct classification of isolated test digits. When our classifier is applied to the recognition of 4958 connected handwritten digit strings (4555 2-digit, 355 3-digit, and 48 4-digit strings) from the NIST Special Database 3 with a dynamic programming approach, it has a correct classification rate of 82.4 percent with a rejection rate of as low as 0.85 percent. Our classifier compares favorably in terms of correct classification rate and robustness with other classifiers that are tested     

Large vocabulary recognition of on-line handwritten cursive words

This paper presents a writer independent system for large vocabulary recognition of on-line handwritten cursive words. The system first uses a filtering module, based on simple letter features, to quickly reduce a large reference dictionary (lexicon) to a more manageable size; the reduced lexicon is subsequently fed to a recognition module. The recognition module uses a temporal representation of the input, instead of a static two-dimensional image, thereby preserving the sequential nature of the data and enabling the use of a Time-Delay Neural Network (TDNN); such networks have been previously successful in the continuous speech recognition domain. Explicit segmentation of the input words into characters is avoided by sequentially presenting the input word representation to the neural network-based recognizer. The outputs of the recognition module are collected and converted into a string of characters that is matched against the reduced lexicon using an extended Damerau-Levenshtein function. Trained on 2,443 unconstrained word images (11 k characters) from 55 writers and using a 21 k lexicon we reached a 97.9% and 82.4% top-5 word recognition rate on a writer-dependent and writer-independent test, respectively     

Handwritten Bangla numeral recognition system and its application to postal automation

A recognition system for handwritten Bangla numerals and its application to automatic letter sorting machine for Bangladesh Post is presented. The system consists of preprocessing, feature extraction, recognition and integration. Based on the theories of principal component analysis (PCA), two novel approaches are proposed for recognizing handwritten Bangla numerals. One is the image reconstruction recognition approach, and the other is the direction feature extraction approach combined with PCA and SVM. By examining the handwritten Bangla numeral data captured from real Bangladesh letters, the experimental results show that our proposed approaches are effective. To meet performance requirements of automatic letter sorting machine, we integrate the results of the two proposed approaches with one conventional PCA approach. It has been found that the recognition result achieved by the integrated system is more reliable than that by one method alone. The average recognition rate, error rate and reliability achieved by the integrated system are 95.05%, 0.93% and 99.03%, respectively. Experiments demonstrate that the integrated system also meets speed requirement.     

A New Hybrid Approach to Handwritten Address Verification

The use of optical character recognition (OCR) has achieved considerable success in the sorting of machine-printed mail. The automatic reading of unconstrained handwritten addresses however, is less successful. This is due to the high error rate caused by the wide variability of handwriting styles and writing implements. This paper describes a strategy for automatic handwritten address reading which integrates a postcode recognition system with a hybrid verification stage. The hybrid verification system seeks to reduce the error rate by correlating the postcode against features extracted and words recognised from the remainder of the handwritten address. Novel use of syntactic features extracted from words has resulted in a significant reduction in the error rate while keeping the recognition rate high. Experimental results on a testset of 1,071 typical Singapore addresses showed a significant improvements from 24.0% error rate, 71.2% correct recognition rate, and 4.8% rejection rate using raw OCR postcode recognition to 0.4% error rate, 65.1% correct recognition rate, and 34.5% rejection rate using the hybrid verification approach. The performance of the approach compares favourably with the currently installed commercial system at Singapore Post, which achieved 0.7% error rate, 47.8% correct recognition rate, and 51.5% rejection rate for 6-digit postcode using the same test data.     

A practical approach for writer-dependent symbol recognition using a writer-independent symbol recognizer

We present a practical technique for using a writer-independent recognition engine to improve the accuracy and speed while reducing the training requirements of a writer-dependent symbol recognizer. Our writer-dependent recognizer uses a set of binary classifiers based on the AdaBoost learning algorithm, one for each possible pairwise symbol comparison. Each classifier consists of a set of weak learners, one of which is based on a writer-independent handwriting recognizer. During online recognition, we also use the n-best list of the writer-independent recognizer to prune the set of possible symbols and thus reduce the number of required binary classifications. In this paper, we describe the geometric and statistical features used in our recognizer and our all-pairs classification algorithm. We also present the results of experiments that quantify the effect incorporating a writer-independent recognition engine into a writer-dependent recognizer has on accuracy, speed, and user training time.     

Unsupervised language model adaptation for handwritten Chinese text recognition

This paper presents an effective approach for unsupervised language model adaptation (LMA) using multiple models in offline recognition of unconstrained handwritten Chinese texts. The domain of the document to recognize is variable and usually unknown a priori, so we use a two-pass recognition strategy with a pre-defined multi-domain language model set. We propose three methods to dynamically generate an adaptive language model to match the text output by first-pass recognition: model selection, model combination and model reconstruction. In model selection, we use the language model with minimum perplexity on the first-pass recognized text. By model combination, we learn the combination weights via minimizing the sum of squared error with both L2-norm and L1-norm regularization. For model reconstruction, we use a group of orthogonal bases to reconstruct a language model with the coefficients learned to match the document to recognize. Moreover, we reduce the storage size of multiple language models using two compression methods of split vector quantization (SVQ) and principal component analysis (PCA). Comprehensive experiments on two public Chinese handwriting databases CASIA-HWDB and HIT-MW show that the proposed unsupervised LMA approach improves the recognition performance impressively, particularly for ancient domain documents with the recognition accuracy improved by 7 percent. Meanwhile, the combination of the two compression methods largely reduces the storage size of language models with little loss of recognition accuracy.     

A writer identification system for on-line whiteboard data

In this paper we address the task of writer identification of on-line handwriting captured from a whiteboard. Different sets of features are extracted from the recorded data and used to train a text and language independent on-line writer identification system. The system is based on Gaussian mixture models (GMMs) which provide a powerful yet simple means of representing the distribution of the features extracted from the handwritten text. The training data of all writers are used to train a universal background model (UBM) from which a client specific model is obtained by adaptation. Different sets of features are described and evaluated in this work. The system is tested using text from 200 different writers. A writer identification rate of 98.56% on the paragraph and of 88.96% on the text line level is achieved.     

A Mixture of Recurrent Neural Networks for Speaker Normalisation

In spite of recent advances in automatic speech recognition, the performance of state-of-the-art speech recognisers fluctuates depending on the speaker. Speaker normalisation aims at the reduction of differences between the acoustic space of a new speaker and the training acoustic space of a given speech recogniser, improving performance. Normalisation is based on an acoustic feature transformation, to be estimated from a small amount of speech signal. This paper introduces a mixture of recurrent neural networks as an effective regression technique to approach the problem. A suitable Vit-erbi-based time alignment procedure is proposed for generating the adaptation set. The mixture is compared with linear regression and single-model connectionist approaches. Speaker-dependent and speaker-independent continuous speech recognition experiments with a large vocabulary, using Hidden Markov Models, are presented. Results show that the mixture improves recognition performance, yielding a 21% relative reduction of the word error rate, i.e. comparable with that obtained with model-adaptation approaches.     

A combination of features for symbol-independent writer identification in old music scores

The aim of writer identification is determining the writer of a piece of handwriting from a set of writers. In this paper, we present an architecture for writer identification in old handwritten music scores. Even though an important amount of music compositions contain handwritten text, the aim of our work is to use only music notation to determine the author. The main contribution is therefore the use of features extracted from graphical alphabets. Our proposal consists in combining the identification results of two different approaches, based on line and textural features. The steps of the ensemble architecture are the following. First of all, the music sheet is preprocessed for removing the staff lines. Then, music lines and texture images are generated for computing line features and textural features. Finally, the classification results are combined for identifying the writer. The proposed method has been tested on a database of old music scores from the seventeenth to nineteenth centuries, achieving a recognition rate of about 92% with 20 writers.     

An omnifont open-vocabulary OCR system for English and Arabic

We present an omnifont, unlimited-vocabulary OCR system for English and Arabic. The system is based on hidden Markov models (HMM), an approach that has proven to be very successful in the area of automatic speech recognition. We focus on two aspects of the OCR system. First, we address the issue of how to perform OCR on omnifont and multi-style data, such as plain and italic, without the need to have a separate model for each style. The amount of training data from each style, which is used to train a single model, becomes an important issue in the face of the conditional independence assumption inherent in the use of HMMs. We demonstrate mathematically and empirically how to allocate training data among the different styles to alleviate this problem. Second, we show how to use a word-based HMM system to perform character recognition with unlimited vocabulary. The method includes the use of a trigram language model on character sequences. Using all these techniques, we have achieved character error rates of 1.1 percent on data from the University of Washington English Document Image Database and 3.3 percent on data from the DARPA Arabic OCR Corpus