Research on Mathematical Formulas Extraction from Chinese Document

A new approach for separating mathematics from usual text is presented. Contrary to the existing researches, it is aiming at Chinese mathematical documents and more oriented toward the segmentation than the recognition, separating the formulas outside and inside the text lines. The approach is composed of Parzen windows and Bayes theorem. Improved Parzen windows is used to extract the isolated formulas from the printed documents and Bayes theorem is used to extract the embedded formulas from the text lines. Experiments show that the combination of the two methods can achieve quite satisfactory result