Arabic Handwritten Character Recognition Using Machine Learning Approaches

Abstract

Character recognition system of Arabic handwritten is one among of the convincing research and analysis tasks going on in light of the fact that every writer has his own writing style. It is the capacity of the PC to understand and recognize the characters of Arabic handwritten automatically. Due to the advancement within the field of technology and science, there is requirement for character recognition of handwritten script in several real-time applications to reduce the effort of human. Several Algorithms of deep learning along with machine learning are improved which may be utilized for the recognition System of character. This manuscript performs the accuracy analysis and algorithms performance measures of CNN (Convolutional Neural Networks), KNN (K-Nearest Neighbor), and SVM (Support Vector Machine). The authors suggested a new model depend on the methods of CNN, KNN, and SVM together. These various machine learning approaches are integrated to develop the recognition system performance. Experimentation has been performed on AHDB and IFN/ENIT data sets. The outcomes of the new model experimental show an excellent performance with higher accuracy.