Optical Formula Extraction Based on Irregularity Degree

Optical formula extraction is considered as an important step of mathematical formula recognition, which can convert scientific papers into their corresponding electronic format. So far little research has been done in this area. This paper proposes an approach of extracting embedded formulas that first invokes a searching algorithm to find the connected components of the input document, calculates the layout feature of every component based on irregularity degree, and then locates the formula symbols according to the features. Finally, several measurements including linking grammar are used to locate the formula areas. The experimental results indicate that the proposed method can obtain favorable results