Computer vision application to automatically recognise handwritten Arabic characters

摘要

This paper aims to use computer vision techniques to automatically recognize (read) off-line Arabic handwriting words. It also concentrates on the feature extraction process, i.e., concentrates on extracting the main geometrical features of each extracted handwritten Arabic character. A complete system able to recognize Arabic-Handwritten characters of only a single writer is proposed and discussed. The system first attempts to remove some of variation in the images that do not affect the identity of the handwritten word (slant correction, slop correction, and baseline estimation). Next, the system codes the skeleton of the word so that feature information about the lines in the skeleton is extracted (segmentation and feature extraction). Each handwritten Arabic character is represented by three segments. Each segment has eleven features. These features representing a character are fed to the classifier (feed-forward error backpropagation neural network in this research) for classification. The number of inputs to the neural network is 37, eleven features for every character segment plus two inputs representing dot(s) if any, plus two representing if the character is connected to other character from left or right side. The features include locating endpoints, junctions, turning points, loops, generating frames (segmentation step), and detecting strokes. These features are then passed on to the recognition system. A 69.7 percent recognition rate has been achieved on the character frames of data.