Poster: Real-time markerless kinect based finger tracking and hand gesture recognition for HCI

Abstract

Hand gestures are intuitive ways to interact with a variety of user interfaces. We developed a real-time finger tracking technique using the Microsoft Kinect as an input device and compared its results with an existing technique that uses the K-curvature algorithm. Our technique calculates feature vectors based on Fourier descriptors of equidistant points chosen on the silhouette of the detected hand and uses template matching to find the best match. Our preliminary results show that our technique performed as well as an existing k-curvature algorithm based finger detection technique.