The Research of ISOETRP Clustering Algorithm on Optical Mathematical Symbols Recognition

A mathematical symbol recognition approach based on ISOETRP Clustering Algorithm is proposed in this paper. Mathematical symbol is one kind of special image symbols, different from ordinary text character in several aspects. For this point, firstly, one special features extraction algorithm is proposed, then two kinds of Classifier are used, one is Minimum Distance Classifier, another is Tree Classifier based ISOETRP clustering Algorithm used to improve the results. Five hundred isolated Mathematical Formulas including 10000 symbols are picked out for testing, and the recognition ratio reaches at 95%. The experimental results show that Tree Classifier is superior to Minimum Distance Classifier, and this method can be applied to practice.