A review of the published research confirms that recognition of printed Arabic Word continues to present challenges. This is specially the case when segmentation is problematic. A word level recognition system is presented here that does not rely on any segmentation or require baseline detection of ascenders and descenders. A Discrete Hidden Markov classifier along with a block-based Discrete Cosine Transform (DCT) is used to construct a novel holistic Arabic printed word recognizer. A balanced database of word-image has been constructed to ensure an even distribution of word samples. The Arabic words are typewritten in five fonts having a size 14 points in a plain style. The system is applied on actual scanned word images with no overlap between the training and testing datasets. Word feature vectors are extracted using block-based DCT. A Hidden Markov Models Toolkit (HTK) is used to construct the recogniser. Vector Quantisation is used to map each feature vector to the closest symbol in the codebook. The output of the system is multiple recognition hypotheses (N-best word lattice). The results are encouraging when compared with other published research in this area achieving on average 97.65% accuracy which is significantly higher than previously published results. A detailed comparison and analysis of the results are presented.

Owing to effectiveness and ease of implementation Sum rule has been widely applied in the biometric research field. Different matcher information has been used as weighting parameters in the weighted Sum rule. In this work, a new parameter has been devised in reducing the genuine/imposter distribution overlap. It is shown that the overlap region width has the best generalization performance as the weighting parameter amongst other commonly used matcher information. Furthermore, it is illustrated that the equal weighted Sum rule can generally perform better than the Equal Error Rate and d-prime weighted Sum rule. The publicly available databases: the NIST-BSSR1 multimodal biometric and Xm2vts score sets have been used.

Pattern analysis and prediction of sensory data is becoming an increasing scientific challenge and a massive economical interest supports the need for better pattern mining techniques. The aim of this paper is to investigate efficient mining of useful information from a sensor network representing an ambient intelligence environment. The goal is to extract and predict behavioral patterns of a person in his/her daily activities by analyzing the time series data representing the behaviour of the occupant, generated using occupancy sensors. There are various techniques available for analysis and prediction of a continuous time series signal. However, the occupancy signal is represented by a binary time series where only discrete values of a signal are available. To build the prediction model, recurrent neural networks are investigated. They are proven to be useful tools to solve the difficulties of the temporal relationships of inputs between observations at different time steps, by maintaining internal states that have memory. In this paper, a special form of recurrent neural network, the so-called Echo State Network (ESN) is used in which discrete values of time series can be well processed. Then, a model developed based on ESN is compared with the most popular recurrent neural networks; namely Back Propagation Through Time (BPTT) and Real Time Recurrent Learning (RTRL). The results showed that ESN provides better prediction results compared with BPTT and RTRL. Using ESN, large datasets are learnt in only few minutes or even seconds. It can be concluded that ESN are efficient and valuable tools in binary time series prediction. The results presented in this paper are based on simulated data generated from a simulator representing a person in a 1 bed room flat.

In this paper, the prediction of the occupancy of different areas in a single-occupant intelligent inhabited environment is addressed It is aimed to deliver a well-being monitoring and assistive environment to support elderly to live independently. A wireless sensor network of motion detection sensors is constructed to collect the required occupancy data. Individual sensory data are combined to form an occupancy time series. Then, a fuzzy predictor model is proposed to model the occupancy time series and the results of this technique are compared with other techniques in time series prediction including auto regressive moving average, adaptive network based fuzzy inference system, as well as transductive neuro-fuzzy inference model with weighted data normalization.

This paper presents the investigations carried out in improving the performance of a holistic cursive word recogniser using approximating splines. The holistic cursive word recogniser and the spline based zone line extraction mechanism are described. A series of tests are carried out using 3600 word images from 18 writes and a 200 word lexicon to compare the performance of the histogram based zone extraction with that of the spline based approach and the results are presented. It is found that the spline based zone extraction significantly improves the recognition results. The future directions of the research for further improvement of recognition performance are presented.

In this paper we present a multimodal interface that employs speech recognition and eye gaze tracking technology for use in data entry tasks. The aim of this work is to compare the usability of this multimodal system against other data entry methods (handwriting, mouse and keyboard and speech only) when carrying out the data entry task of filling a form. Discussions regarding the relationships between efficiency, effectiveness, ergonomic quality, hedonic quality, naturalness, familiarity and users preference are presented. The experimental results show that the majority of the users prefer using the proposed eye and speech system compared to the other form-filling methods even though such a method is neither the fastest nor the most accurate.

In the work carried out earlier [1][2], it was found that an eye gaze and speech enabled interface was the most preferred form of data entry method when compared to other methods such as mouse and keyboard, handwriting and speech only. It was also found that several non-native United Kingdom (UK) English speaking speakers did not prefer the eye gaze and speech system due to the low success rate caused by the inaccuracy of the speech recognition component. Hence in order to increase the usability of the eye gaze and speech data entry system for these users, error recovery methods are required. In this paper we present three different multimodal interfaces that employ the use of speech recognition and eye gaze tracking within a virtual keypad style interface to allow for the use of error recovery (re-speak with keypad, spelling with keypad and re-speak and spelling with keypad). Experiments show that through the use of this virtual keypad interface, an accuracy gain of 10.92% during first attempt and 6.20% during re-speak by non-native speakers in ambiguous fields (initials, surnames, city and alphabets) can be achieved [3]. The aim of this work is to investigate whether the usability of the eye gaze and speech system can be improved through one of these three multimodal blended multimodal error recovery methods.

An extensive literature survey on automated assessment and handwriting recognition has shown that no work has been done in addressing the area of assessment of handwritten exam scripts. We therefore introduce the novel concept of applying image extraction and cursive script recognition (CSR) techniques to the area of automated assessment. We demonstrate the potential for using a holistic CSR engine as the input process for a system capable of automatically scoring handwritten responses to multi-choice questions. This innovative system utilises the constrained nature of simple multiple choice questions to enhance the recognition rate of the handwritten response. Fifty writers were chosen to answer eight multiple choice questions and results show that the system yields an average 83% CSR word accuracy, which enables the system to score over 54% of all response with 99% confidence.

This paper describes an independent handwriting style classifier that has been designed to select the best recognizer for a given style of writing. For this purpose a definition of handwriting legibility has been defined and a method has been implemented that can predict this legibility. The technique consists of two phases. In the feature extraction phase, a set of 16 features is extracted from the image contour. These features have been selected from amongst a set of pre-recognition features as those features that contribute the most (95%) to a discriminant between legible and illegible words. In the classification phase, a Probability Neural Network based on Bayesian decision is introduced to predict the legibility of unknown handwriting using a Parzen method to estimate a class conditional density function from the available training data.

There is a pressing need for very large collections of training data if cursive script recognition research is to become a commercial success. However, the construction and proofing of such databases is extremely laborious. This paper describes the use of a tool which facilitates this process, and which has enabled the authors to construct a useful collection of data which is available to the research community. The structure of the database is an important consideration in making it useful to other researchers, and this is described along with a range of access functions and tools.

The advances in Optical Character Recognition (OCR) technology over the past decade have enabled the development of many automatic document-processing systems capable of 99% correct recognition on printed text. However, similar advances in Cursive Script Recognition (CSR) technology have not been forthcoming due, principally, to the vast variability of human handwriting. This paper investigates a method by which the more reliable OCR technology can be used to improve the CSR performance in a form processing application. A novel method is proposed to link handwriting data to contextual cue words that have been automatically obtained from an OCR process. This information is then used to select appropriate 'focused' lexicons to achieve better CSR results. The method was tested on 30 forms that were filled by 10 different writers. The experimental results together with a comparison to the base line recognition performance are presented.

Colour has long been viewed as one of the unnecessary features in any form processing system, due not only to the large storage requirement and computational cost its inclusion imposes but also to the complexities of hue, chroma and brightness variation. However, as technology has advanced and computing costs have reduced, the processing of documents in colour has now become practical. This paper describes a prototype form extraction system that utilises colour information to help facilitate data extraction from a form. Blank forms are first automatically analysed to obtain their layout, colour and statistical information. The filled data is then extracted from the filled forms using techniques based upon the colour characteristic changes that have occurred with respect to the blank form. The improved performance of the proposed method has been verified by comparing the processing time, data extraction precision and recall rate of the proposed system to that of an archetypal black and white form extraction system.

Presents an efficient technique for the classification of off-line handwritten words into upper and lower case using principal components (PC). The technique consists of two phases. For each word, in the feature extraction phase, first the boundary points of the word are extracted, then 26 features, including global, local, regional and dominant features, are extracted using the contour information. In the classification phase, a discriminant function based on the PC, adapted by our system, is introduced to integrate the extracted features and classify words into upper and lower case. Experimental results show that the system achieves 83% correct word case classification for about 2240 test words randomly selected from a data set of 3226 words obtained from 12 writers.

This paper presents the research carried out in producing a whole recognizor for cursive handwritten words in facsimile images. Two sets of handwritten data samples are collected and converted into facsimile images. The first set comprises approximately 1600 word images from 8 writers and is used for development purposes. The second set consists of approximately 2000 word images from 10 writers. This set is used for testing only. The algorithms for extraction of holistic features namely, vertical bars, holes and cups used in the recognizor are described. A series of test are carried out and the results are presented using a 200 word lexicon. The holistic recognizor produced 62% top rank and 82% in top 5 alternatives. When a lexicon of 1000 words was used these values reduced to 49% and 70% respectively. The future directions of the research for improvement of recognition rate are proposed. It is envisaged that definition of further features would improve the overall accuracy.

Automatic analysis and recognition of documents is a complex task. Substantial research effort has therefore been spent addressing sub-tasks of manageable size. The solutions, however, have to be integrated into a comprehensive engine. Following the idea of a toolbox a software architecture has been developed that enables various tools to be combined into a network of cooperating recognition algorithms. Distributing the tools on a number of processors improves overall performance by parallel execution. The configuration of the tools as well as the connections between them can easily be modified even at run-time. Simple, interactive access to the various parameters and intermediate results of an algorithm aids the development and evaluation of experimental systems. The modular approach also facilitates extension of existing systems by the integration of novel methods. DART, the Distributed Asynchronous Recognition Toolbox architecture, realises these features. DART has been successfully used for recognition of handwriting and poor quality facsimiles.

The recognition of unconstrained handwriting has to cope with the ambiguity and variability of cursive script. Preprocessing techniques are often applied to on-line data before representing the script as basic primitives, resulting in the propagation of errors introduced during pre-processing. This paper therefore combines pre-processing of the data (i.e. tangential smoothing) and encoding into primitives (Partial Strokes) in a single step. Finding the correct character at the correct place (i.e. letter spotting) is the main problem in non-holistic recognition approaches. Many cursive letters are composed of common shapes of varying complexity that can in turn consist of other subshapes. In this paper, we present a production rule system using Hierarchical Fuzzy Inference in order to exploit this hierarchical property of cursive script. Shapes of increasing complexity are found on a page of handwriting until letters are finally spotted. Zoning is then applied to verify their vertical position. The performance of letter spotting is compared with an alternative method.

A method for the recognition of poor quality documents containing touching characters is presented. The method is based on extraction of independent and robust features of each object of a sample word, where objects consist of single letters or of several touching ones. Thus avoiding letter segmentation the method eliminates errors frequently introduced in segmentation based approaches. Features are attributed by their position and extent in order to facilitate discrimination between different classes of objects. A method for automatic construction of a comprehensive database is presented. From a given dictionary every possible letter combination is obtained and the images of the artificially touching letters created. These images are subjected to noise and their features extracted. For recognition, alternatives for each object are found based on the database. Object alternatives are then combined into valid word alternatives using lexicon lookup. It has been observed that the developed method is effective for the recognition of poor quality documents.

This paper presents an novel approach for tackling problems associated with flexibility of dynamic structures. A number of solutions to this problem have been developed by innovative combination of fuzzy logic and neural networks inexact algorithms. A spring-mounted pen is used in the experiments to emulate the deviation of an end-effector caused by flexibility. A pre- and post-processing vision-based machine control system is developed. Comparing the desired pattern and the actual output, the intelligent machine console can automatically make appropriate compensation through online self-learning process. Various experimental results indicate that using the algorithms developed the system can compensate for flexibility and produce excellent results, much better than human operators.

The paper describes a holistic recognizer developed for use in a hybrid recognition system. The recognizer uses information about the word shape. As this information is strongly related to word zoning, care is taken to avoid limitations resulting from the inaccuracy of zone detection. The recognizer uses a very simple set of features and a fuzzy set based pattern matching technique. This aims to increase its robustness, but also causes problems with disambiguation of the results. A verification mechanism, using letter alternatives as compound features, is introduced. The letter alternatives are obtained from a segmentation based recognizer coexisting in the hybrid system. The holistic recognizer is found capable of outperforming the segmentation based one, despite the remaining disambiguation problems. When working together in a hybrid system, the results are significantly higher than those of the individual recognizers. Recognition results are reported and compared.

The paper concentrates on the combination of results of multiple recognizers at the word level. Two approaches are presented: word list merging and linear combination. Word list merging requires no knowledge about the individual recognizers. The linear combination is an attempt to exploit the information about characteristics of individual recognizers. This appears more complex than in the case of combination of results at the character level. Recognition of words is influenced by more factors, which can independently affect the recognition process. Characterisation of recognizers, used for word level combination, is more complex and requires more than a simple consideration of recognition success and failure. The concept of handwriting data characterisation is defined. A number of handwriting characteristics are extracted and used to guide the combination process. The choice of characteristics is made in the context of recognition methods used. No attempt at general characterisation of handwriting is made. The relationship between handwriting characteristics and recognition results is observed and used to obtain characteristics of individual recognizers. Results of the two combination methods are reported and compared with another frequently used method for results combination, the Borda count.

Cursive script recognition is commonly based on finding letters within a word and recognizing them separately. The segmentation process is ambiguous and difficult. A method which combines word segmentation and letter recognition with lexical look-up in order to cope with segmentation ambiguity is presented. Words are first segmented into small elements which are then put together using a database of their possible combinations to produce alternative segmentations. Letter recognition is performed on each letter candidate and lexical look-up is applied, interactively, to prune illegal word recognition results. Lexical look-up is used to postulate word endings for partially recognized words. This provides the means of catering for words with sloppy endings, some misspellings and recovering from some recognition errors. An online cursive script recognition system, based on the above method, is described and evaluated.>