Gesture recognition system for NOOR eMashaf application: Concepts and architecture

摘要

This paper describes the concepts and architecture of a gesture recognition system to be implemented as a navigation component for NOOR eMashaf Android-based mobile application for reading and learning the Holy Quran. It is based on motion detection methods that extract a user's hand profile in motion under dynamic lighting conditions and against natural background. The proposed system is being developed into three main phases: detection, tracking and recognition. In the first phase, The detection of the hand, in a natural scene, is captured by a front-facing camera, then the tracking phase where the hand's motion is traced, finally the recognition phase should identify the trajectory path of the hand. At a later stage, the captured path is compared with a model database that contains pre-defined gestures, once the gestures are recognized the system will activate the corresponding action. The proposed hand motion detection method proved to be very effective under varying lighting conditions and against natural and complex backgrounds. Thus it is adequate enough to be implemented and integrated with mobile applications in general, and with NOOR eMashaf application, in our case, that enables the control and execution, by the proposed hand gestures, of a specific action or event of the mobile application. Promising results has been obtained so far, at the initial stages of the project.