Zone-Wise Segmentation and Lexicon-Driven Recognition for Printed Myanmar Characters

Abstract

This paper presents a new segmentation and recognition algorithms for Myanmar script inputted from offline printed images. Zone segmentation considers horizontal and vertical zones; it is applied to segment letters according to their roles such as primary or peripheral characters. In doing so, statistical and structural features of segmented characters are explored and exploited in recognition process. Hidden Markov model is used for recognition of primary characters while Kohonen self-organization map is used for peripheral characters. The recognized characters by each model are then combined, and finally are recognized by k-nearest neighbors algorithm with the help of lexicon is composed of all common Myanmar characters. Our OCR system for Myanmar characters tested on a dataset that approximately contains 7560 compounded characters. From the results, our system achieves higher significant results both segmentation and recognition compared to the other contemporary Myanmar OCR’s approaches.