Using image recognition to facilitate intuitive input

Abstract

Intuitive interaction with computer systems can be enhanced by letting the user manipulate physical objects. In this paper we describe a novel input system, allowing the user to specify translations and rotations by manipulating cardboard shapes. A camera records the shapes' configuration. Some optimized image recognition algorithms, based on the Hough transform, are used to collect positional and orientational data concerning the cardboard shapes. This information is translated to manipulations on virtual objects in the application.