A Preclassification Method for Handwritten Chinese Character Recognition Via Fuzzy Rules and Seart Neural Net

Abstract

In this paper, a method of character preclassification for handwritten Chinese character recognition is proposed. Since the number of Chinese characters is very large (at least 5401s for daily use), we employ two stages to reduce the candidates of an input character. In stage I, we extract the first set of primitive features from handwritten Chinese characters and use fuzzy rules to create four preclassification groups. The purpose in stage I is to reduce the candidates roughly. In stage II, we extract the second set of primitive features from handwritten Chinese characters and then use the Supervised Extended ART (SEART) as the classifier to generate preclassification classes for each preclassification group created in stage I. Since the number of characters in each preclassification class is smaller than that in the whole character set, the problem becomes simpler. In order to evaluate the proposed preclassification system, we use 605 Chinese character categories in the textbooks of elementary school as our training and testing data. The database used is HCCRBASE (provided by CCL, ITRI, Taiwan). In samples 1–100, we select the even samples as the training set, and the odd samples as the testing set. The characters of the testing set can be distributed into correct preclassification classes at a rate of 98.11%.