The use of orthographic and lexical information for handwriting recognition

Wells, C.J., 1992. The use of orthographic and lexical information for handwriting recognition. PhD, Nottingham Trent University.

Preview

Text 10290157.pdf - Published version Download (28MB) | Preview

Abstract

This thesis details the research work undertaken by the author from July 1987 to May 1992 concerning the automatic recognition of handwriting by computer. The emphasis has been the development of algorithms and data structures which facilitate the use of orthographic and lexical information to resolve the ambiguity present in handwriting recognition systems.

Good recognition performance is ultimately linked not only to the capacity of the system to extract and compare good feature sets, but also to the integration of context and knowledge in the different processing stages. Hence the current study forms the first part of a contextual recognition system. It describes a number of levels of analysis for handwriting recognition, which begin with the application of orthographic information. Letters do not combine arbitrarily to form words, so letter string combinations produced by a pattern recogniser can be checked for acceptability. Such a process is known as a lexical check and requires a list of words or lexicon against which to compare alternative candidate strings.

A list of words can be acquired from a standard dictionary. However the words must be stored in a suitable data structure which can easily be searched to check the existence of candidate letter strings. This structure should preferably be searchable in real time and have only modest memory requirements so that it could be part of a recognition system on a personal computer. These aspects are considered within this thesis. The advantages and disadvantages of a number of methods are introduced and compared. The lexical analysis system described can interface to higher level linguistic constraints which aid the recognition process.

In the future it is hoped that the computer could be a "notepad" style portable. The interface to such computers should mimic conventional pen and paper instead of using a keyboard. Hence efficient and reliable handwriting recognition will necessarily form an important part of this new technology.

Item Type:	Thesis
Creators:	Wells, C.J.
Date:	1992
ISBN:	9781369324068
Identifiers:	Number Type PQ10290157 Other
Rights:	This copy of the thesis has been supplied on condition that anyone who consults it is understood to recognise that its copyright rests with its author and that no quotation from the thesis and no information derived from it may be published without the author’s prior consent.
Divisions:	Schools > School of Science and Technology
Record created by:	Linda Sullivan
Date Added:	10 Nov 2020 15:44
Last Modified:	05 Oct 2023 09:47
URI:	https://irep.ntu.ac.uk/id/eprint/41592

Actions (login required)

Edit View

Views

Downloads