Cursive Script Recognition : A Sentence Level Recognition

We describe the cursive script recognition module of a cheque processing system currently under development at the Centre for Pattern Recognition and Machine Intelligence (CENPARMI). Common systems perform recognition either on a character by character basis, or on a word level. The present study investigates the recognition at a higher level of abstraction, at the sentence level. Our computational theory is based on a psychological model of the reading process of a fast reader. In this paper, we discuss the processing of the legal amount written on cheques. The preprocessing and feature extraction modules as well as the word classiier are presented. Preliminary results are not only promising, but they also support our computational theory.