Recognition of mathematical expressions

A machine-implemented method for recognizing a handwritten mathematical expression, the machine-implemented method comprising:

• partitioning a plurality of strokes of the handwritten mathematical expression, included in a region having at least one non-terminal object, into one or more groups of two regions;
• repeating the act of partitioning on ones of regions included in the one or more groups of two regions having a non-terminal object;
• using a first recognizer and a second recognizer to calculate first scores and second scores, respectively, of chart entries representing grammar objects formed from one or more strokes included in the regions;
• calculating first converted scores and second converted scores from the first scores and the second scores, respectively, each of the converted scores being calculated according to:
• converted_score
• =
• T
• ×
• dir
• ×
• score
• -
• m
• d
• ,
• where converted score is the converted score, T is a parameter obtained during training for the first recognizer or the second recognizer, correspondingly, score is a score from the first recognizer or the second recognizer, correspondingly, m is a mean score with respect to sample scores of the first recognizer or sample scores of the second recognizer, correspondingly, d is a sample deviation of the sample scores of the first recognizer or the sample scores of the second recognizer, correspondingly, and dir is a direction constant with respect to the first recognizer or the second recognizer, correspondingly;
• determining a respective converted score for each of the regions based on highest scores selected from corresponding first converted scores and corresponding second converted scores;
• and
• recognizing the plurality of strokes of the mathematical expression based on finding highest converted scores of partitioning of the plurality of strokes, wherein
• the machine-implemented method is performed by at least one processor.