Structural analysis of handwritten mathematical expressions through fuzzy parsing

Including mathematical expressions in documents can be a tiresome and difficult process. A recognition system for handwritten mathematical expressions would greatly simplify the task, enabling users to enter expressions in an intuitive manner. In this paper a powerful yet efficient approach to handwritten mathematical expression recognition is presented. Our approach is built upon traditional shiftreduce parsing methods which have linear complexity, thus providing the necessary syntax checking and also the basis for an efficient system. Fuzzy logic is introduced to the basic parsing framework, which is appropriate given the ambiguous nature of symbol identities and spatial relationships in handwritten expressions. Multiple parses are explored whenever strong ambiguities arise, and the most likely expression tree is selected as the result.