Handwritten Mixed Numeral Classification System

Abstract

Digital revolution implies an essential role in every field. It is widely used in banks, shops, government agencies, industries, etc. which provides more efforts to doing work, also it reduces chances of mistakes and gives quick result. Pattern recognition is one of the interesting branches of the digital revolution. Basically, it is used to detect a proper set of pattern to provide proper functioning of any devices. Pattern recognition is based on OCR system which stands for Optical Character Recognition. OCR uses its preset data and compares it with any new data and if matches are correct it gives output. Dealing with printed data format is quite easier because it has fixed set of patterns, where handwritten data is different person by person. So it is quite tedious because there are numbers of irregularities available in handwritten data. A handwritten mixed numeral classification system based on OCR. Here we are using different numerals from different languages and generate a system which can deal with handwritten data presented in many more languages. There is two stage of this systems 1) training stage and 2) testing stage. We are going through several image processing steps at both stages. Also, we can deal with the different classifier and finally provide a system which can deal with many more languages and gives good output with high accuracy.