Vision enabled computer games : implementation of a Virtual Badminton Trainer

Abstract

Vision is a critical component in human-to-human communication. The richness of face-to-face communication is far greater than that of voice or text. But if we consider way humans and computers are interacting visually, the use of artificial or mechanical means is evident. Though, the human predominantly reacts visually, the computer hardly does so, except for producing its output. This research aims at bridging this gap by imparting visual human computer interaction. Even though the material described in this thesis concentrate mainly on computer games, the proposed methods are applicable to many other application areas, too. A camera mounted in front of a player will capture the motion of the player. Information extracted from this video stream can then be used to bring the human player into the virtual playing field. The virtual shuttlecock will be generated virtually according to the parameters defined in the computer. The virtual playing field will be projected on a large screen in front of the human player. The racket detection algorithm is based on colour and shape descriptors. After extracting the racket it will be analyzed to identify its orientation. Then it will be combined with previous badminton racket positions, to find out its velocity in order to detect its collision and its outcome with the virtual shuttlecock.