A real-time gesture recognition implementation on SoC development platform

This paper proposes a real-time gesture recognition algorithm, describes the implementation of it on a SoC development platform, and shows its application. In our design, we adopt the vision-based approach to identify the gestures. The images are retrieved from webcam at the rate of 30 fps, and then go through the pre-processing, and noise cancelation phases. After that, we use template matching to identify what the gestures are. Finally, we combine with the application of digital photo frame and demonstrate using our gesture recognition system in real-time to control the digital photo frame which is operated by touch-screen previously.