Handwritten Character Recognition by Fuzzy Learning Vector Quantization

Abstract

Learning Vector Quantization (LVQ) proposed by T.Kohonen is a simple algorithm and also has a strong pattern recognition power. It is possible for Fuzzy Learning Vector Quantization (FLVQ) to construct networks with a higher ability to extract characteristics of pattern data than that of LVQ. This paper proposes a pattern matching method based on the learning algorithm of FLVQ for handwritten Japanese character (KANJI) recognition. The authors make it clear through recognition experiments that the proposed method is useful even for complex structured characters but still has some difficulties.