Handwriting Arabic Words Recognition Based on Structural Features
Abstract
Handwriting recognition technology is the ability of a computer to recognize characters, words and other symbols that have been written by hand in natural handwriting. This study presents a method for recognition of Handwritten Arabic Words (HAW) through expanding in the way of structural features extraction by relying on geometrical information (straight lines, loops, points, and curve). The input to the system is binary images written by hand by number of people. The features are to convert the image from two dimensional into one dimensional as a victor that is to be used as a signature for the image the experiments have been conducted on a database of a thousand words representing names of a hundred Libyan cities at a rate of ten patterns for each city. Classification of the words was dependence on Artificial Neural Networks (ANN) of Multiple Layers Perceptron (MLP) type. Wherein half of the words were used to train the network and the other half to test the network. The ratio of recognition was 80.4 %.