Analysis of human gestures in the 3D space to control multimedia interfaces

The recognition of movements and poses of the body is a subject of fundamental importance to create simple, new and effective methods of natural interaction between man and machine. The complete understanding of human gesticulation is the crucial base for creating intuitive and easy to use natural interfaces. The main purpose of this paper is to examine the state of the art, illustrating techniques of particular interest for the recognition of stroke, trajectories and 3D or 2D gesture. Moreover, here the implemented methods and techniques used in order to recognize three different types of inputs: simple poses of the hands, fast swipe and complex gestures in 3D space are described. Motion analysis has been made using a depth camera and a skeleton detection algorithm. The complex gestures are compared using an improvement of the technique called Dynamic Time Warping (DTW). This technique was analyzed in a very accurate way, in order to understand which implementation would lead to an effective, reliable and fast recognition of gestures. The results and the implementated methodologies have been accurately exposed and adequately compared.