On-line recognition handwritten mathematical symbols

A new method of basic element ordering is proposed and the number of standard symbols is reduced by extracting three primary features of mathematical symbols, namely, basic element vector, relative positions between basic elements and basic element length vector.

Abstract

The paper aims at online recognition of handwritten mathematical symbols. It analyses the structures of 94 opt used mathematical symbols and concludes that all of them consist of 10 basic elements. It proposes a new method of basic element ordering and reduces the number of standard symbols by extracting three primary features of mathematical symbols, namely, basic element vector, relative positions between basic elements and basic element length vector. The traditional dynamic programming method is improved by means of classifying roughly 94 mathematical symbols, considering matching and unmatching value, adding geometric restraints and solving matching problem, through improved Kohn-Munkres algorithm. Correctness rate reaches 90.52%, incorrectness rate 5.03% and refusal rate 4.45%.