Object Parameter Extraction for Control in Vision-Based Tabletop Systems

TitleObject Parameter Extraction for Control in Vision-Based Tabletop Systems
Publication TypeMaster Thesis
Year of Publication2008
AuthorsEstapé, O.
preprint/postprint documentfiles/publications/Octavi-Estape-Master-Thesis.pdf
Abstract

reacTIVision is an open source computer vision framework for recognition and tracking of fiducials. It finds its main application in development of table-based tangible user interfaces (TUI) and multi-touch interactive surfaces (tabletops). This Master Thesis aims to improve the current reacTIVision system. It reviews the state of the art of the algorithms that can be used for parameter extraction from generic objects, focusing in the algorithms that can be used for control and real-time interaction. Among all shape description techniques, three different and complementary techniques have been implemented: some simple global descriptors give us the position and orientation of the shape with a rough classification, the polygon simplification can be used for user feedback and polygon recognition, and finally a simplification of the Medial Axis Transform (skeleton) is better suited for deformable objects.

intranet