Image Based Equation Solver

Abstract

This work presents an Android based mobile application that takes Images of handwritten Mathematical expressions as input, recognize the symbols and operators in the expression and solves/evaluates it. The developed application is capable of solving single line expression including arithmetic operators. The application mainly relies on image processing and pattern classification techniques to recognize the handwritten symbols. The different modules of the project have been implemented using the OpenCV library (for image processing operations), Tesseract engine (for recognition of text) and Exp4j library (for solving the Mathematical expressions). An image of handwritten expression fed to the system is binarized and connected components are extracted. A collection of handwritten digits, operators and variable symbols extracted from a set of documents are used to train the Tesseract engine. During evaluation phase, the query expression image is segmented into components and each component is fed to the trained recognition engine which outputs the class label of the component. The expression converted from image to text is presented to the user for validation or edition. Finally, the validated expression is evaluated and the final result is presented to the user. The system evaluated on a variety of handwritten expressions realized promising performance. iv