A static hand gesture recognition system for real time mobile device monitoring

The mobile human-computer interaction is aiming to facilitate communication with the mobile devices. The hand gesture recognition is considered as the most important alternative that can be deployed because it is natural, intuitive and easy to use. In this paper, we proposed our system of static hand gesture recognition to control mobile devices based only on a real time video streaming capture from the front-facing camera of the device. Our proposed system is based on the skin colour algorithm and face subtraction to detect hand area. The features derived from contour extraction, convex hull detection, convexity defects extraction and the palm center detection are used on SVM classifier for the recognition step. This system, running on an Android tablet, achieves 96,8% of recognition rate that proves the robustness of our system. We applied the parallelism techniques to ensure the respect of real time constraint and realise a 23 fps.