In the recent few years, there was a concentrated search on Arabic Optical Character Recognition (OCR), especially the recognition of scanned, offline, machine-printed documents. However, Arabic OCR consequences are dissatisfying and are still a developed research area. Finding the best feature extraction techniques and selecting an appropriate classification algorithm lead to supreme recognition accuracy and low computational overhead. This paper presents a new Arabic OCR model by integrating both of Genetic Algorithm (GA) and the Fuzzy K-Nearest Neighbor classifier (F-KNN) in a unified framework to enhance the identification accuracy. GA is utilized as a feature selection algorithm that has better convergence and spread of solutions with candid variation preservation mechanism. The F-KNN algorithm is more appropriate to classify ambiguous or uncertain data objects in the sense that every object belongs to all classes with different degrees of membership. The suggested model semantically fuses bio-inspired based feature vectors with fuzzy KNN classifier to build accurate membership function for each class. Experimental results compared to other approaches revealed the effectiveness of the suggested model and demonstrated that the feature selection approach increased the identification accuracy process.

No abstract