A Multi-Camera Framework for Interactive Videogames

Abstract

Modern days, the traditional input devices for computer games are keyboard, mouse and joystick. But there also exists other hardware to interact with games, for example a microphone or a camera. Most of them are still in research state, but they are finding their way to the commercial world. The Eye Toy of Sony Playstation 2 is a good example for this. The purpose of this thesis is to describe such an alternative input for games by using two or more cameras. Because of this, the player must be able to interact intuitively in 3D computer games without physical contact with any gaming device so that everyone can play those games. Using multiple cameras for computer games has to meet with two properties, it has to be cheap and it has to be interactive. In the past there has been a lot of research for creating cheap and efficient hardware for this purpose, but modern cameras are often cheap enough. So the first property is not a problem anymore because for example web cams are very low cost. The second demand has been researched in the computer vision and image processing area a lot and there are many ways to do this, each one has his advantages and disadvantages. It is a large research domain and there are no specific formulations of how computer vision problems should be solved. This work will present a collection of techniques and algorithms capable of extracting useful information from captured images in real time and how this information can be used for interaction in computer games. This can all be done with a standard single processor computer and no specialized hardware.