Metric Vision

This is a home page dealing with issues involved in "Metric Vision", i.e., gaining quantative information on the position, orientation and dimension of objects from images. The page has been implemented and will be maintained by Paul O'Leary and Matthew Harker, Institut for Automation, University of Leoben, A-8700 Leoben, Austria, automation.unileoben.ac.at.

The present focus of the work presented here is on signal processing and fitting geometric models to features, which have been extracted from images. The identification of various projections, e.g. homographies, is also dealt with since they are required during calibration processes.

At this stage (Auguat 2005) we are at the start of this project and many of the pages are still empty. The starting point for this project a library of functions and methods for the fitting of geometric objects to data, it could be called a collection of "Geometric Recipes". At this point in time MATLAB® implementations are presented and Maple® derivations are given where available. In some cases documentation is provided in PDF format. It is planned to extend the library to invlude LabVIEW® code.

This site has the following structure:

Introduction, gives lecture notes and tutorials and the methods used in the implementation of the library;

Geometry, contains the library of geometric methods for metric vision;

Signal Processing, will contain information on signal processing techniques used;

Hardware, will contain information of hardware used, inparticular on IEC 61131 compatible connection of industrial I/O.

Publications, a list of publications related to this work.