A novel approach for detecting and recognizing mathematical symbols

Abstract

Extraction and recognition of mathematical characters and equations for documented images is important for many applications, artificially intelligent systems in particular to store or analyze mathematical data. It is vital that efficient and robust approaches are utilized in this sector. Recognition methodologies that are widely used at present work very well for nonmathematical data, but for mathematical data, the results are somewhat unsatisfying. Moreover, precision is important while extracting mathematical texts. In our thesis we have studied a number of text detection algorithms and used image segmentation methodologies, template matching, SVM (Support Vector Machine) to detect and recognize mathematical symbols from a document image.