Addis Ababa University Libraries Electronic Thesis and Dissertations: AAU-ETD! >
Faculty of Informatics >
Thesis - Computer Science >
Please use this identifier to cite or link to this item:
|Title: ||Writer Independent Online Handwriting Recognition for Ethiopic Characters|
|Authors: ||Daniel, Negussie|
|Advisors: ||Dr. Solomon Atnafu|
|Copyright: ||2006 |
|Date Added: ||29-Aug-2008 |
|Publisher: ||Addis Ababa University|
|Abstract: ||A prototype for a writer independent online handwriting recognition system for Ethiopic characters has been designed.
Nowadays, online handwriting systems are one of the most famous ways of interacting with computers. Despite this fact, the absence of localized pen computing applications for Ethiopic script called for the development and implementation of a writer-independent online handwriting recognition system for Ethiopic characters.
Therefore, in this study the first prototype to ever be developed fro a writer independent online handwriting recognition system for Ethiopic characters is demonstrated.
The prototype incorporates a preprocessing and classification module. The design of this prototype has concentrated on the thirty four first order Ethiopic characters with the vision of expanding the data corpus to include all Ethiopic characters and numerals. The data was collected from thirty randomly chosen individuals using a Wacom pen and tablet digitizer. Neuroscript MovaAlyzer was the software used to sample data points. The matching algorithm is Dynamic Time Warping. This algorithm calculated the distances between characters in the database and has yielded a recognition accuracy of 86.9%.
With the expansion of the data corpus this recognition rate has promising prospects ahead.|
|Description: ||A THESIS SUBMITTED TO THE SCHOOL OF GRADUATE STUDIES OF
ADDIS ABABA UNIVERSITY IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF MASTERS OF SCIENCE IN
|Appears in:||Thesis - Computer Science|
Items in the AAUL Digital Library are protected by copyright, with all rights reserved, unless otherwise indicated.