Abstract:Given the ubiquity of handwriting and mathematical content in human transactions, machine recognition of handwritten mathematical text and symbols has become a domain of great practical scope and significance. Recognition of mathematical expression (ME) has remained a challenging and emerging research domain, with mathematical symbol recognition (MSR) as a requisite step in the entire recognition process. Many variations in writing styles and existing dissimilarities among the wide range of symbols and recurring characters make the recognition tasks strenuous even for Optical Character Recognition. The past decade has witnessed the emergence of recognition techniques and the peaking interest of several researchers in this evolving domain. In light of the current research status associated with recognizing handwritten math symbols, a systematic review of the literature seems timely. This article seeks to provide a complete systematic analysis of recognition techniques, models, datasets, sub-stages, accuracy metrics, and accuracy details in an extracted form as described in the literature. A systematic literature review conducted in this study includes pragmatic studies until the year 2021, and the analysis reveals Support Vector Machine (SVM) to be the most dominating recognition technique and symbol recognition rate to be most frequently deployed accuracy measure and other interesting results in terms of segmentation, feature extraction and datasets involved are vividly represented. The statistics of mathematical symbols-related papers are shown, and open problems are identified for more advanced research. Our study focused on the key points of earlier research, present work, and the future direction of MSR.

Language Models for Handwritten Short Message Services

A Unified Multilingual Handwriting Recognition System using multigrams sub-lexical units

Offline recognition of unconstrained handwritten texts using HMMs and statistical language models

Representing Online Handwriting for Recognition in Large Vision-Language Models

Fully Convolutional Networks for Handwriting Recognition

Modeling the lexical morphology of Western handwritten signatures

Integrating Language Model in Handwritten Chinese Text Recognition.

Digitizing Handwriting with a Sensor Pen: A Writer-Independent Recognizer

Towards an IMU-based Pen Online Handwriting Recognizer

Data Driven Design Of An Ann/Hmm System For On-Line Unconstrained Handwritten Character Recognition

TextCaps : Handwritten Character Recognition with Very Small Datasets

Retrieval-based Language Model Adaptation for Handwritten Chinese Text Recognition

A Lexicon and Depth-wise Separable Convolution Based Handwritten Text Recognition System

Improving Accuracy and Explainability of Online Handwriting Recognition

Revisiting N-Gram Models: Their Impact in Modern Neural Networks for Handwritten Text Recognition

Vision-Language Model Based Handwriting Verification

Introducing a new high-resolution handwritten digits data set with writer characteristics

A novel connectionist system for unconstrained handwriting recognition

Handwritten Word Recognition using Deep Learning Approach: A Novel Way of Generating Handwritten Words

Handwritten Text Recognition System using Machine Learning

Machine learning models for mathematical symbol recognition: A stem to stern literature analysis