The concept of key characters in a cursively handwritten word image is intr
oduced and a method for extracting the key characters is presented. Key cha
racters capture the unambiguous parts of the cursive words that can be reli
ably segmented and recognized. We propose a method for lexicon reduction us
ing key characters in conjunction with a word-length estimation. (C) 1999 E
lsevier Science B.V. All rights reserved.