Handwritten Polynomial Equation Solver Using Convolutional Neural Network

Abstract

More the number of people more is the diversity in handwritten charactersand hence task of recognizing characters becomes intricate. In this paper, we proposed a Customized Convolutional neural network (CCNN) model which has made recognition more precise than the existing recognitionalgorithms. Customized ConvolutionalNeuralNetwork(CCNN)extractsfeaturesautomaticallyandhandleslarge amount of data thus resulting in better accuracy. Classification and Segmentation of characters is a challenge on based of machine vision system. The study looks at polynomial equations written by hand as well as dataset. Dataset of 10 digits, 4 operators and 6 variables is used for performance evaluation of proposedmodel.This input image, we apply projection horizontally for the equation to be segmented. After this, each segmented character is sent into our neural network model i.e. CNN, for character indexing. A character string that looks similar the earlier inputequation. Every approach is required for solution. Ourmodeldoesn’trestrictanyn-dimensionaldataandproducesbetter results. Customized Convolutional Neural Network (CCNN) extracts features automatically and handles a large amount of data thus resulting into a better accuracy. Finally, the strategy used is effective as described in results of experiment.This paper explains the methodology and architecture of our CNN model which can be further referred to in future and used for solving other pattern and recognition problems.