Off-line handwritten arabic characterssegmentation using slant- tolerant segment features (STSF)
Loading...
Date
2007
Authors
Shubair, A. Abdullah
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The main theme of this research is the off-line handwritten Arabic characters
segmentation. A successful handwritten Arabic character recognition system improves
interactivity between the human and the computers. Building successful Arabic
character recognition system cannot be fulfilled without solving the segmentation
problem. One of the major problems is the slant of the words and the dissimilarity in
length and slope of the junction line between the handwritten characters. The challenge
is to extract the morphological features and to find the junction line. The foremost
contribution of this research is the new segmentation algorithm for handwritten Arabic
characters using the Slant-Tolerant Segments Feature (STSF). STSF is a string of '+'
or '-' signs and represents the strokes of the word characters. The testing of the STSF
algorithm was conducted using a newly created database. The new database is called
Arabic Handwritten Database/Universiti Sains Malaysia (AHD/USM). The Arabic
handwritten database consists of 12300 words written by 82 different writers aged from
5 to more than 45 years. The experiments reported 90.12% correctness rate. The
research also contributes in facilitating the study of the Arabic characters segmentation
problem by creating new categorization system for the segmentation methods. It
categorizes the segmentation methods into two approaches: Junction-Seeking
Approach (JSA) and Recognize-Segment Approach (RSA).
Description
Master
Keywords
Language , Slant tolerant