Improved Topological Fiducial Tracking in the reacTIVision System

This paper describes reacTIVision: a camera based two dimensional fiducial (marker) tracking system developed for the reacTable*, a table based tangible musical instrument. Key features of reacTIVision include: (1) the ability to track a large number of fiducials with faster than real-time performance and (2) fiducial size may be varied depending on the number of distinct fiducial identities required. We describe recent advances in our implementation of topologybased fiducial recognition, including a generalised method for accurately computing fiducial location and orientation. A method of graph naming known as left heavy depth sequences is applied to the identification of topologically distinct fiducials. Also discussed is our approach to generating fiducial images for the system, in which we employ evolutionary computation to produce compact fiducials with specific geometric properties.