Digit recognition using different features extraction methods

Abstract

The Numeral features extraction consists of transforming the image into an attribute vector, which contains a set of discriminated characteristics for recognition, and also reducing the amount of information supplied to the system. Several characteristic extractions methods have been proposed in the literature. These characteristics can be digital or symbolic. Mainly, we distinguish two approaches, statistical and structural. In this paper, we are interested in a comparative study of these four methods: profile projection, zoning, cavities and freeman chain code. Digit recognition is carried out in this work through k nearest neighbors. We evaluated our scheme on handwritten samples of the MNIST database and we have achieved very promising results.