Heuristics approchto optical character recognition

Abstract

A principal goal of the research upon which this report is based is to develop a methodology for recognition of characters of any font and size. With referenced to the Indian context, recognition of different scripts, such as Devanagari, Bengali, Telugu and Tamil have been made. But no serious effort has been made to develop character recognition scheme for Gurmukhi script that recognize printed documents of any font and size.The challenge is to develop methods for character recognition that remove as much as possible the font and size restrictions while maintaining overall rates close to those of the limited front commercial machines. This report describes a serious effort towards developing much a system. I tested the alorithm on Gurmukhi script.Recogniation of Gurmukhi Script characters without performing the step of pre prescribe is proposed in this thesis. This method recognizes characters of Gurmukhi script without pre-processing the step of pre-processing which usually takes a lot of time. An advantage of the method is that all the features are extracted in one horizontal and one vertical scan. The effectiveness of the approach combining statistical and structural approaches was observed. With this proposed method as recognition rate of 89.41.has been achieved.