Design of an Effective Preprocessing Approach for Offline Handwritten Images

Abstract

Handwritten pattern recognition involves conversion of scanned images of handwritten patterns into a computer processable form. To recognize handwritten patterns is an easy and trivial task for human beings, but for a machine it is a cumbersome and a difficult task due to high variations in the shape of characters and writing style. Although complicated to train, yet machines can be useful in providing solution to the recognition problem. They save time and money and eliminate the requirement of execution of repetitive tasks by humans. In order to have better recognition the image should be properly pre-processed. Pre-processing reduces and eliminates noise and irregularities. The present paper focuses on different approaches to pre-processing and an insight to general methodology for the recognition process. General TermsPattern Recognition