Library Automation and Digital Archive
LONTAR
Fakultas Ilmu Komputer
Universitas Indonesia

Pencarian Sederhana

Find Similar Add to Favorite

Call Number SEM-348
Collection Type Indeks Artikel prosiding/Sem
Title Multi stroke recognition based on rubine gesture classification using fuzzy matching and frequency, 87-100
Author Priadhana Edi Kresnha, M. Rahmat Widyanto;
Publisher Proceedings international seminar information technology (isit) 2009 it for pride and wealth of nation november 25th,2009 lumire hotel and pascasarjana stmik nusa mandiri building/menara salemba
Subject
Location
Lokasi : Perpustakaan Fakultas Ilmu Komputer
Nomor Panggil ID Koleksi Status
SEM-348 TERSEDIA
Tidak ada review pada koleksi ini: 45154
Stroke recognition has been extensively studied to improve interaction between human and computer. usually, human has to follow the rule and standard from computer application to create drawing. like UML, diagram, etc. human can not freely express and draw as much as he wants to complete the drawing. by recognizing human drawing, computer will be able to analyze what user wants and what user will create, even the computer will be able to repair the user's drawing. using rubine gesture classifier, a stroke is defined and included into a class. then y using rubine gesture recognition, a stroke can be detected and identified, into which class it should be classified based on characteristic it has. there are 13 characteristic that rubine method extracted to recognize a stroke. unfortunately, rubin method is only effective for single stroke. to be able to recognize multiple stroke, several methods have to be implemented. fuzzy matching and frequency are used to detect multiple strokes. frequency is used to memorize user habits, computer may recognize and estimate what drwaing the user wants to make. fuzzy matching is used to deal with the fuzziness of the direction of the stroke and the direction of the next stroke based on the current stroke. SVM is used as well to classify and identify the strokes. experiment shows that rubine classifier is effective to classify and detect stroke, and combining it with fuzzy matching and frequency of the stroke drawn by the user, multi stroke can be recognized well.