Handwritten Digit Recognition Using Neural Network with Gabor Filter for Information Fusion

Abstract

Because of several aspects such as variability in individual’s personal writing style, handwriting recognition is the most difficult area to master. Despite the massive amount of research and development that has gone into it, it has yet to tackle all of the commercially important and conceptually intriguing concerns. Many document processing and evaluation systems that use digital image processing techniques need handwritten digit recognition. In many pattern recognition applications, document processing and document analysis are becoming more ubiquitous. The aim of this paper is to propose a system which will take a digital image as an input and will automatically detect and display the text present in it. This will save the time and efforts and will reduce the chances of involvement of human error in the system. This system can be further used in many applications like reading postal addresses, bank check amount, forms, etc. The paper used a Gabor filter bank for extracting features from a digit image. This reduction in preprocessing overhead brings up the performance of the system in prediction of digits. The prediction of this system is almost real time with very high accuracy.