Shape recognition using fuzzy string-matching technique

Abstract

Object recognition is a very important task in industrial applications. Attributed string matching is a well-known technique for pattern matching. The present paper proposes a fuzzy string-matching approach for two-dimensional object recognition. The fuzzy numbers are used to represent the edit costs. Therefore, the edit distances are also presented as fuzzy numbers. The attributed string-matching problem is then equivalent to a fuzzy shortest path problem. The edit distance between two shapes is presented as a fuzzy number. By ranking the fuzzy edit distances, the input shape is classified as the reference shape that has the minimum fuzzy edit distance. The experimental results show that the proposed method can effectively recognize two-dimensional objects.