Ch. Leung et L. Sze, FEATURE-SELECTION IN THE RECOGNITION OF HANDWRITTEN CHINESE CHARACTERS, Engineering applications of artificial intelligence, 10(5), 1997, pp. 495-502
A method is proposed here to extract appropriate features for the reco
gnition of handwritten Chinese characters. The features represent the
lengths, positions and directions of the character strokes. In additio
n, two approaches (Karhunen-Loeve and stroke density analyses) have be
en used to analyze the information content of Chinese characters, from
which a cost-effective, non-uniform, and two-dimensional sampling sch
eme for feature extraction has been derived. The resulting scheme extr
acts more samples from regions of higher information content. A recogn
ition system that combines all the proposed approaches was built, and
experiments were performed on the 500 most frequently used Chinese cha
racter classes, with 20,000 handwritten samples. Results indicated tha
t the proposed methods are useful. (C) 1997 Elsevier Science Ltd. All
rights reserved.