Binary printed documents are used in a wide variety of applications, for example contracts, wills... more Binary printed documents are used in a wide variety of applications, for example contracts, wills, recommendation letters, etc. Such documents are intrinsically valuable but as a results may be subject to attacks by forgers who would make deliberate changes to the documents, for example changing a name or a date in a contract. However, during transmission (e.g. printing, in the mail, when being scanned), they will be subject to noise, for example spurious marks or errors in printing may occur. This thesis presents a methodology for use in the verification of such documents. The methods used seek to reject deliberately forged documents while at the same time not rejecting those which have only been altered by the addition of noise. The thesis demonstrates that current techniques used for other types of document are either unacceptably expensive or intrinsically unsuited to binary documents. The methods developed have been tested on a range of deliberately forged documents as well as on valid documents and are shown to be robust. Limitations in the methods have been identified and solutions to the problems are proposed.
ABSTRACT This paper introduces a novel approach for dynamic structuring of contextual lattices. I... more ABSTRACT This paper introduces a novel approach for dynamic structuring of contextual lattices. It is anticipated that the approach can be applied to improve the accuracy of word-segmentation patterns in autonomous text recognition systems. A multi-level hierarchical structure of lattices is used to implement the algorithm, and the approach can be applied in a generic manner to other pattern recognition problems. We apply a top-down structural model in parallel with a constrained probabilistic model and intelligent distributed searching paradigm. This paradigm is based on the integration between probabilistic bi-grams and adaptive intelligent swarm-based agent search to identify the most likely sentence structures. The searching paradigm allows the exploitation of positive feedback as a search mechanism and, consequently, makes the model amenable to parallel implementation. The distributed intelligence of the proposed approach enables the dynamic structuring of contextual lattices and has proved to scale well with large lattice sizes. Moreover, we believe that the proposed architecture solves the ill-conditioned nature of most pattern recognition problems that lies in the effect of noise in the segmentation phase. To verify the developed Swarm-based Intelligent Search Algorithm (SISA), a simulation study was conducted on a set of variable size scripts. The proposed paradigm proved to be efficient in identifying the most highly segmented patterns and also returned good decisions concerning lower probability segments enabling further re- segmentations and re-combinations to take place. The paper is the first to apply the intelligent swarm-based paradigm for the identification of optimal segmented patterns in contextual recognition models. The algorithm is compared with other algorithms for the same problem, and the computational results demonstrate that the proposed approach is very efficient and robust for large-scale statistical contextual-lattice structures.
7th International Conference on Image Processing and its Applications, 1999
ABSTRACT The recognition of cursive handwriting remains a challenging an open problem, especially... more ABSTRACT The recognition of cursive handwriting remains a challenging an open problem, especially for the case of off-line recognition from a scanned image. The authors research on the recognition of handprinted characters has achieved some success as reported in Elliman et al. (1991). In attempting to recognise increasingly careless handprint it becomes necessary to segment touching characters, and to join separated ones. The distinction between cursive writing and handprint is a fuzzy one with no clear dividing line, and it became evident that our methods should generalise across this spectrum if they were to be successful for general document reading. Familiarity with the literature of the field led the authors to believe that an extension of their approach to handprint could be useful for the recognition of cursive script, especially in the more challenging case of off-line recognition. Every year at Christmas time one of the authors is faced with a strong challenge to his own ability to recognise handwriting, in the form of a letter from his Auntie Pauline. An extract from the 1998 letter was scanned and is taken as an example for testing the approach
The annual construction of an Exam Timetable is a common problem for all institutions of higher e... more The annual construction of an Exam Timetable is a common problem for all institutions of higher education. Quite often it is done by hand or with the limited help of a simple administra- tion system and usually involves taking the previous year's timetable and modifying it so it will work for the new year. Many British institutions are now introducing
Object recognition supported by user interaction for service robots, 2002
Abstract Hidden Markov models (HMM) have been used with some success in recognizing printed Arabi... more 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 ...
Proceedings of ICSIPNN '94. International Conference on Speech, Image Processing and Neural Networks, 1994
... Cellular automata (CA) were conceived in the late forties by John Von Neumann to provide a re... more ... Cellular automata (CA) were conceived in the late forties by John Von Neumann to provide a realistic model for ... This paper shows that the system makes use of a genetic algorithm to successfully find the correct function for ... 0-7803-1865-X/94/$3.00 0 1994 IEEE 753 ISSII"N'94 - ...
Proceedings of Sixth International Conference on Document Analysis and Recognition, 2001
ABSTRACT There is a pressing need for very large collections of training data if cursive script r... more ABSTRACT 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
IEE Colloquium on Document Image Processing and Multimedia Environments, 1995
... the algorithm. Baird[2] suggests a skew detec-tion algorithm with involves abstracting each c... more ... the algorithm. Baird[2] suggests a skew detec-tion algorithm with involves abstracting each character to reference point from which energy points are calculated through a range of conceivable angles. This technique seemed ...
IEE Third European Workshop on Handwriting Analysis and Recognition, 1998
ABSTRACT The author has developed software in the past that segments hand-printed characters usin... more ABSTRACT The author has developed software in the past that segments hand-printed characters using simple heuristics, normalises them, extracts features and recognises them using a quadratic discriminant function. He concluded that correct segmentation was the area that needed major research effort. If this could be done correctly then the recognition results would be reasonably impressive using existing software. He has developed software for producing outlines, and then strokes (or vectors) from scanned images. The original application was the capture of engineering drawings. The outlining is a crack-following approach. A sequence of mergelets forms a word. Finding the best sequence of mergelets is the formulation of the segmentation problem
http://tawny.cs.nott.ac.uk/ttg/index.html Abstract. This paper discusses a series of recombinatio... more http://tawny.cs.nott.ac.uk/ttg/index.html Abstract. This paper discusses a series of recombination operators for the timetabling problem. These operators act upon a direct representation of the timetable and maintain the property of feasibility . That is that there are no conflicts and no overfilled rooms. V arious approaches to solving the timetabling problem using evolutionary computing methods are first compared. The recombination operators are then presented and various alternatives for incorporating heuristic knowledge in the search are described. Finally, results are presented comparing the operators on a real timetabling problem.
Binary printed documents are used in a wide variety of applications, for example contracts, wills... more Binary printed documents are used in a wide variety of applications, for example contracts, wills, recommendation letters, etc. Such documents are intrinsically valuable but as a results may be subject to attacks by forgers who would make deliberate changes to the documents, for example changing a name or a date in a contract. However, during transmission (e.g. printing, in the mail, when being scanned), they will be subject to noise, for example spurious marks or errors in printing may occur. This thesis presents a methodology for use in the verification of such documents. The methods used seek to reject deliberately forged documents while at the same time not rejecting those which have only been altered by the addition of noise. The thesis demonstrates that current techniques used for other types of document are either unacceptably expensive or intrinsically unsuited to binary documents. The methods developed have been tested on a range of deliberately forged documents as well as on valid documents and are shown to be robust. Limitations in the methods have been identified and solutions to the problems are proposed.
ABSTRACT This paper introduces a novel approach for dynamic structuring of contextual lattices. I... more ABSTRACT This paper introduces a novel approach for dynamic structuring of contextual lattices. It is anticipated that the approach can be applied to improve the accuracy of word-segmentation patterns in autonomous text recognition systems. A multi-level hierarchical structure of lattices is used to implement the algorithm, and the approach can be applied in a generic manner to other pattern recognition problems. We apply a top-down structural model in parallel with a constrained probabilistic model and intelligent distributed searching paradigm. This paradigm is based on the integration between probabilistic bi-grams and adaptive intelligent swarm-based agent search to identify the most likely sentence structures. The searching paradigm allows the exploitation of positive feedback as a search mechanism and, consequently, makes the model amenable to parallel implementation. The distributed intelligence of the proposed approach enables the dynamic structuring of contextual lattices and has proved to scale well with large lattice sizes. Moreover, we believe that the proposed architecture solves the ill-conditioned nature of most pattern recognition problems that lies in the effect of noise in the segmentation phase. To verify the developed Swarm-based Intelligent Search Algorithm (SISA), a simulation study was conducted on a set of variable size scripts. The proposed paradigm proved to be efficient in identifying the most highly segmented patterns and also returned good decisions concerning lower probability segments enabling further re- segmentations and re-combinations to take place. The paper is the first to apply the intelligent swarm-based paradigm for the identification of optimal segmented patterns in contextual recognition models. The algorithm is compared with other algorithms for the same problem, and the computational results demonstrate that the proposed approach is very efficient and robust for large-scale statistical contextual-lattice structures.
7th International Conference on Image Processing and its Applications, 1999
ABSTRACT The recognition of cursive handwriting remains a challenging an open problem, especially... more ABSTRACT The recognition of cursive handwriting remains a challenging an open problem, especially for the case of off-line recognition from a scanned image. The authors research on the recognition of handprinted characters has achieved some success as reported in Elliman et al. (1991). In attempting to recognise increasingly careless handprint it becomes necessary to segment touching characters, and to join separated ones. The distinction between cursive writing and handprint is a fuzzy one with no clear dividing line, and it became evident that our methods should generalise across this spectrum if they were to be successful for general document reading. Familiarity with the literature of the field led the authors to believe that an extension of their approach to handprint could be useful for the recognition of cursive script, especially in the more challenging case of off-line recognition. Every year at Christmas time one of the authors is faced with a strong challenge to his own ability to recognise handwriting, in the form of a letter from his Auntie Pauline. An extract from the 1998 letter was scanned and is taken as an example for testing the approach
The annual construction of an Exam Timetable is a common problem for all institutions of higher e... more The annual construction of an Exam Timetable is a common problem for all institutions of higher education. Quite often it is done by hand or with the limited help of a simple administra- tion system and usually involves taking the previous year's timetable and modifying it so it will work for the new year. Many British institutions are now introducing
Object recognition supported by user interaction for service robots, 2002
Abstract Hidden Markov models (HMM) have been used with some success in recognizing printed Arabi... more 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 ...
Proceedings of ICSIPNN '94. International Conference on Speech, Image Processing and Neural Networks, 1994
... Cellular automata (CA) were conceived in the late forties by John Von Neumann to provide a re... more ... Cellular automata (CA) were conceived in the late forties by John Von Neumann to provide a realistic model for ... This paper shows that the system makes use of a genetic algorithm to successfully find the correct function for ... 0-7803-1865-X/94/$3.00 0 1994 IEEE 753 ISSII"N'94 - ...
Proceedings of Sixth International Conference on Document Analysis and Recognition, 2001
ABSTRACT There is a pressing need for very large collections of training data if cursive script r... more ABSTRACT 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
IEE Colloquium on Document Image Processing and Multimedia Environments, 1995
... the algorithm. Baird[2] suggests a skew detec-tion algorithm with involves abstracting each c... more ... the algorithm. Baird[2] suggests a skew detec-tion algorithm with involves abstracting each character to reference point from which energy points are calculated through a range of conceivable angles. This technique seemed ...
IEE Third European Workshop on Handwriting Analysis and Recognition, 1998
ABSTRACT The author has developed software in the past that segments hand-printed characters usin... more ABSTRACT The author has developed software in the past that segments hand-printed characters using simple heuristics, normalises them, extracts features and recognises them using a quadratic discriminant function. He concluded that correct segmentation was the area that needed major research effort. If this could be done correctly then the recognition results would be reasonably impressive using existing software. He has developed software for producing outlines, and then strokes (or vectors) from scanned images. The original application was the capture of engineering drawings. The outlining is a crack-following approach. A sequence of mergelets forms a word. Finding the best sequence of mergelets is the formulation of the segmentation problem
http://tawny.cs.nott.ac.uk/ttg/index.html Abstract. This paper discusses a series of recombinatio... more http://tawny.cs.nott.ac.uk/ttg/index.html Abstract. This paper discusses a series of recombination operators for the timetabling problem. These operators act upon a direct representation of the timetable and maintain the property of feasibility . That is that there are no conflicts and no overfilled rooms. V arious approaches to solving the timetabling problem using evolutionary computing methods are first compared. The recombination operators are then presented and various alternatives for incorporating heuristic knowledge in the search are described. Finally, results are presented comparing the operators on a real timetabling problem.
Uploads
Papers by David Elliman