A Feature Extraction Method for Arabic Offline Handwritten Recognition System Using Naïve Bayes Classifier

Abstract

Handwriting recognition in the Arabic language is considered one of the most challenging problems and the accuracies in recognizing still need more enhancements due to the Arabic character's nature, cursive writing, style, and size of writing in contrast to working with other languages. In this paper, we propose a system for Arabic Offline Handwritten Character Recognition based on Naïve Bayes classifier (NB). Extraction features preceded by divided the image of character into three horizontal and vertical zones and 3x3 zones in one and two dimensions respectively, then classified by Naïve Bayes. The performance of the system proposes evaluated by using the benchmark CENPARMI database reached up to 97.05% accuracy rate. Experimental results confirm a high enhancement inaccuracy rate in comparison with other Arabic Optical Character Recognition systems.

Arabic Optical Character Recognition: A Review

“…( 1) shows the formula of the Bayes theorem. Abdalkafor et al [91] proposed a handwritten Arabic OCR approach, the character's image is divided into 3 × 3 zones for feature extraction, and then Naïve Bayes is used for classification. The approach was evaluated on the CENPARMI dataset and achieved a 97.05% recognition rate.…”