Abstract:
The Gesture Controlled Application (GCA) has been developed in C# with the aid of Kinect for
windows SDK 1.6 and EmguCV (Wrapper of OpenCV in .NET). The application provides two
types of gesture recognition, Hand gesture recognition and Body gesture recognition. In the former,
we have developed an algorithm that recognizes different hand orientations irrespective of the body
joints using EmguCV to extract hand features including contours, convex hulls and convexity
defects. In the latter, we have used DTW (Dynamic Time Warping) algorithm for the classification
of body gestures using already trained examples and upper body joints information at runtime. The
salient feature of our application is that it can be integrated with many of the windows apps that use
mouse and keyboard. Moreover a user can customize gestures according to his/her will. In other
words, the user can train the Kinect to recognize only a specific set of gestures trained by the user
himself which can prove to be a key application for security purposes in home/office automated
systems using Kinect.
