Offline Handwritten Mathematical Expression Recognition using Convolutional Neural Network

Abstract

Offline Handwritten Mathematical Expression recognition is recognizing the Handwritten Mathematical Expression(HME) that are written on paper, paper is scanned and image is passed to the recognition system. This has caught the interest of most of the researchers and a lot of them are working on this topic and have used various classifiers. In past, Convolutional Neural Network also called CNN has gained high performance in recognizing the patterns. We propose an idea to recognize offline HME using CNN for classification. The steps involved are data collection of handwritten mathematical expression and symbols. Next, preprocessing steps are performed on the collected data. Segmentation of HME into individual symbols is done. To train CNN classifier, Handwritten Mathematical symbols (HMS) are used. The individual segmented symbols are then sent to the CNN classifiers to recognize which class they belong.