Arabic handwritten characters recognition system, towards improving its accuracy
Ahmed Talat Sahlol, Mohamed Elhoseny, Esraa Elhariri, Aboul Ella Hassanien · 2017
Although the extensive work towards building Optical Character Recognition systems(OCR) for Arabic handwritten characters, the unlimited variation and different writing styles of each character make building such these systems a big research challenge. In Arabic alphabetic system, each character has different forms (three or four) depending on its position in a word. In this paper, a handwritten character recognition system was proposed. The proposed system is implemented using a set of well-known optimizers, Bat Algorithm (BAT), Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Grey Wolf Optimization (GWO) algorithm. The proposed system was tested by well-known classifiers to test the efficiency; linear discriminant analysis, support vector machines and random forest. Among all of them, GWO greatly improves the classification accuracy and time efficiency. Compared to the state-of-the-art methods, the optimized feature sets were efficient than the whole feature set in terms of accuracy as well as time consumption.