Abstract
Online handwriting recognition, OHR, has gained a widespread use in everyday life. In some scripts such as Farsi and Arabic, additional strokes are written after the main stroke. These delayed strokes include dots and small signs. In this paper, the delayed strokes effect was studied from two points of views: subword modeling and lexicon reduction. The model of a subword was made of concatenating the main body model and the delayed strokes models. Hidden Markov model, HMM, was employed as a classifier. The delayed strokes of an input subword were additionally exploited to reduce the lexicon size. Our proposed method was tested on TMU-OFS dataset, including 1000 online Farsi subwords, and a recognition rate of 85.2% was achieved.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
