Arabic document image classification using neural networks

Abstract EN

The Neural Network Arabic Document Image Classification System (NNADICS) is an adaptive Arabic document classifier.

By training NNADICS on a number of different document image types, NNADICS behaves as a multiple classifier, since it is capable for distinguishing between multiple document image types.

NNADICS is designed, built, tested and evaluated.

After training NNADICS a document image is applied to the system for classification.

Before that the document image is scanned, pre-processed and binarized, and then applied to NNADICS to classify its contents to text, geometric, or photographic image type.

NNADICS achieved an average of a 86% recognition rate as it is clearly demonstrated