Novel two-stage approach for translation, rotation and scale invariant character recognition

Abstract

This paper presents a new invariant character recognition (ICR) algorithm called as Symmetry Axis based Feature Extraction and Recognition (SAFER) for recognizing translation, rotation and scale invariant optical characters. Unlike most previous ICR algorithms, SAFER puts emphasis on both simplicity and generalization ability for recognizing invariant characters. SAFER uses the axis of symmetry of character and a simple radial coding to extract invariant features from character. The use of ANN as a classifier in SAFER increases its generalization ability. SAFER has been tested on two wellknown fonts Arial and Tahoma that are widely used for representing English characters. These fonts have been subject of many studies in ICR. The experimental results show that SAFER can produce high recognition rate with good generalization ability.