Numerals Recognition System using Feature Extraction and Minimum Distance Technique

Abstract

Numerals Recognition is that strategy planned to determine the digit class from the text image after segmentation process, by convert the identifying and recognizing characters gating from entering images to ASCII cod or any other proportional machine editable structure. OCR systems are generally appropriate for many tasks such as data entry in business documents (e.g. passport, check and so on.), recognized automatic number plate, multi choice examination etc. Numerals recognition is getting increasingly more consideration since 10 years ago because of its wide extent and scope of uses. Arabic number recognition system is proposed in this paper. The proposed system is designed using MATLAB 2017 programming language. This paper intends to develop a method to recognize numbers. The image is read by the program; the next process is feature extraction. Six features were extracted from each number. Using those extracted features, with database created for recognition purpose, the number image is classified by minimum distance technique, where a tested image is compared with the features of every number in database. The result has shown that all the numbers were correctly recognized.