AN OFF-LINE CURSIVE HANDWRITING RECOGNITION SYSTEM

This paper describes a complete system for the recognition of off-line handwriting. Preprocessing techniques are described, including segmen tation and normalization of word images to give invariance to scale, s lant, slope and stroke thickness. Representation of the image is discu ssed and the skeleton and stroke features used are described. A recurr ent neural network is used to estimate probabilities for the character s represented in the skeleton. The operation of the hidden Markov mode l that calculates the best word in the lexicon is also described. Issu es of vocabulary choice, rejection, and out-of-vocabulary word recogni tion are discussed.