An efficient image thresholding method for Arabic handwriting recognition system

Abstract EN

Image preprocessing has assumed an essential part of handwriting recognition system.

The main primary stage of the image preprocessing is thresholding.

An effective thresholding method is based on Fuzzy C-Means clustering (FCM) for Arabic Handwriting Recognition system (AHR) has been proposed in this paper.

Since thresholding stage in AHR is imperative to reduce the dimensionality of image to remove the undesirable information (noise) then increase the processing speed of the AHR system.

The algorithm is performing by feeding the intensity of the pixel value of the image pixels into the FCM clustering algorithm.

Exploratory results with artificial and real life images show that the proposed method gives better accuracy and good efficiency than the current methods.