Stroke Extraction for Offline Handwritten Mathematical Expression Recognition

Abstract

Offline handwritten mathematical expression recognition is often considered much harder than its online counterpart due to the absence of temporal information and the presence of background noise. In order to take advantage of the more developed techniques on online recognition and save resources, an oversegmentation approach is proposed to recover strokes from a textual bitmap image automatically. The proposed algorithm first break down the skeleton of a binarized image into junctions and segments, then segments are merged to form strokes, finally the ordering is determined by recursive projection and topological sort.Given a state-of-art online handwritten mathematical expression recognition system, the proposed procedure correctly recognized 58.22%, 65.65% and 65.05% of the offline formulas rendered from CROHME 2014, 2016 and 2019 respectively. Therefore, the effectiveness of stroke extraction to offline recognition is justified.

A comparative study on methods and tools for handwritten mathematical expression recognition

“…For commercial tool, we have chosen MathPix due to its accuracy, and robusteness.. Since TAP is an online method, we have used the stroke extractor described in [3], allowing to work with TAP in an off-line mode. The models trained and provided by the authors for each network were used.…”

ICDAR 2019 CROHME + TFD: Competition on Recognition of Handwritten Mathematical Expressions and Typeset Formula Detection

“…Fourth, strokes are sorted using a recursive X-Y cut and topological sort. Finally, after extracted strokes are recognized by MyScript Interactive Ink (version 1.3) using a customized grammar and DPI of 576, exported MathMLs are converted to CROHME's conventions and then converted to symLGs [11].…”

Two Decades of Bengali Handwritten Digit Recognition: A Survey

“…However, this area has a long way to go compared to the present state of mathematical expression recognition on other scripts [190]. Advanced techniques like stroke extraction [191] are found to be useful on the publicly available mathematical expression dataset CROHME [192]. Curating such benchmark datasets for Bengali handwritten mathematical expression recognition along with a robust pipeline might be useful for many practical applications.…”