SEGMENTASI KARAKTER TULISAN TANGAN DENGAN FILTER OPTIMAL

Widodo, Suryarini and Malenda, Sariffudin and ETP, Lussiana (2008) SEGMENTASI KARAKTER TULISAN TANGAN DENGAN FILTER OPTIMAL. Seminar Nasional Sistem & Teknologi Informasi (SNASTI) 2008.

[img]
Preview
Text
segmentasi karakter tulisan tangan dengan filter optimal_UG.pdf - Submitted Version

Download (13Mb) | Preview

Abstract

Handwriting has been studied for a long time, because of its widespread use as a means for human-to -human communication. The variability of handwriting, due to differences in cultures, writer styles, alphabets or fonts, make it an interesting problem in the field of pattern recognition. For cursive writing, segmentation is necessary within strokes since several characters can be made with one stroke. Cursive script recognition is difficult because several characters can be written with a single stroke. Owing to the difficulty of this problem, there have been many serious efforts toward obtaining a solution. In this paper, we propose and experimentally study an on-line handwriting segmentation system, we pay attention to the preprocessing stage. This step is necessary because our system considers on-line handwritten specimen as a signal with dynamic information, that can be split into components, which are in turn divided to elemental movements, called strokes. The objective of segmentation is to divide complex handwriting pieces into simpler ones, in order to reduce input pattern variability and thus simplify the classifier structure. In this work initial data are isolated characters composed of one or more components, which are to be segmented into strokes. There are 25 stroke to segment 26 letters in English. We propose optimal filter to smoothing the signal of the character. The segmentation algorithm is based on the maximum and minimum speed analysis of the pen's movement. The percentage of success depends on the quality of the handwriting. The vibration of the hand may affect handwriting quality. In the case of26 letters, the success rate is 73%

Item Type: Article
Uncontrolled Keywords: Smoothing; Optimal Filter Segmentation;
Subjects: A General Works > AI Indexes (General)
Divisions: Fakultas Teknologi Industri > Program Studi Teknik Informatika
Depositing User: Mr Reza Chandra
Date Deposited: 28 Feb 2014 10:03
Last Modified: 28 Feb 2014 10:03
URI: http://repository.gunadarma.ac.id/id/eprint/1211

Actions (login required)

View Item View Item