LTEX Generation from Printed Equations

This article describes a system that takes a photograph of a printed equation and produces a LTEX code representation. The process uses adaptive thresholding with mean filtering, morphological edge smoothing, and the Hough transform for image binarization and skew correction. The project uses Hu invariant moments and circular topology to match characters against a database of characters. The algorithm then assembles the appropriate LTEX code from the detected characters. This system is able to detect 93.6% of characters in ideal images and 86.2% of characters in real-world photographs when combining two different skew methods (74%-79%).