Smartphone Arabic Signboards Images Reading

In this paper, we present the integration of preprocessing, segmentation and Arabic words recognition system. The obtained system is adapted to be executed by smartphone as an application to help pilgrims (HAJEEJ) from different nationalities to automatically read Arabic signboard images taken by their mobiles and recognize their location. The proposed system involves three main approaches i) an existing approach based on Mathematical Morphology (MM) preprocessing, ii) an Outer Isothetic Cover (OIC) segmentation approach and ii) a Transparent Neural Network (TNN) recognition approach. Note that the proposed system, is a smart one in the way it provides the adequate rules of the next pilgrimage step according to HAJEEJ current position. Hence for such smart system, it would be more fruitful to be suitable not only for desk/lab top machines but mainly for any mobile devices. The proposed system is applied on real database mobile images of specific HAJJ places to evaluate recognition rate, time and memory consuming which are necessary for mobile applications.