New method of features extraction for numeral recognition

Authors

  • Ouafae El MELHAOUI LETSER Laboratory, Faculty of Sciences (FSO), Mohammed First University (UMP), Oujda, 60000, Morocco
  • Mohammed BARBOUCHA esearch Center, High Studies of Engineering School, EHEI, Oujda, Morocco
  • Soukaina BENCHAOU MATSI Laboratory, High School of Technology (ESTO), Mohammed First University (UMP), Oujda, 60000, Morocco

Keywords:

eatures extraction, Numeral recognition, k nearest neighbors, multilayer perceptron.

Abstract

This paper proposes a new method of features extraction for handwritten, printed and isolated numeral recognition. It consists of scanning the image row by row, for each row the positions of two first black white transitions in columns were detected, so the first attribute vector is defined from these positions, after, the image was scanned column by column, the positions of two first transitions in row was defined, so the second vector is defined from these positions, so the concatenation of the two vectors determined the attribute vector of the particular numeral. Numeral recognition is carried out in this work through k nearest neighbors and multilayer perceptron. The recognition rate obtained by the proposed system is improved indicating that the numeral features extracted contain more details.

Downloads

Published

28-03-2022

Issue

Section

Articles