APPLICATION OF FUZZY THEORY TO HANDWRITTEN CHARACTER-RECOGNITION

There has been much research in recent decades on character recognitio n methods, and some methods have already been put into practical use. There are many unresolved problems, however, with respect to handwritt en character recognition as composed with printed character recognitio n. The authors considered discriminant functions, which constitute the most important part of a character recognition method. As a result of considering problems of conventional statistical discriminant functio ns, the authors propose applying the fuzzy theory to discriminant func tions. The so-called fuzzy discriminant function is capable of represe nting a data distribution in a more flexible manner because it consist s of membership functions on the principal axes of learning samples. T he authors conducted recognition experiments for handwritten character s with two types of membership functions. In one type the membership v alues are directly tuned based on human experiences; in the other they are derived from histograms or statistical data. With the former memb ership function, the recognition rate of 99.0 percent is achieved for 'numeric' characters from the handwritten alphanumeric data base ETL6, and with the latter, the rate of 96.0 percent for 'hiragana' characte rs from handwritten educational 'kanji' data base ETL8. This result pr oves the effectiveness of the fuzzy discriminant function. It also ind icates that a dynamic combination of human experiences and statistical techniques is a key to practical systems.