Mechanical Engineering The University of Adelaide Australia

FireWire Vision Tools

A simple MATLAB interface to ”FireWire Digital Cameras” (such as UniBrain's fire-i) and the Color Machine Vision algorithms “CMVison”.




The toolbox “FireWire Vision Tools” comprises a small collection of MATLAB drivers for the “UniBrain fire-i” firewire camera (http://www.unibrain.com) or any other firewire camera. Both CMEX drivers as well as S-Functions have been developed to provide high-level access to the underlying hardware as well as to the algorithms of the Color Machine Vision project “CMVision”. A user can thus work with a stream of live images from the MATLAB command line (CMEX interface) or by using intuitive Simulink blocks (S-Functions). Two cameras can be daisy chained to allow for simple stereoscopic image processing. “FireWire Vision Tools” can be downloaded here. A concise user manual is available, as well as a short explanation on how to train the camera for colour vision.

Some of the toolbox commands include rudimentary image processing capabilities. It is possible to classify image contents according to a programmable colour range and/or object size criteria. The vision algorithms can be used to determine the centroid coordinates of a detected object. The automatic tracking of coloured objects is particularly useful in machine vision based robotics applications such as the “RoboCup challenge” and other camera controlled multi-agent systems.

Internally, the drivers gain access to the video capture card through calls to functions provided by the 1394 Digital Camera Driver developed at the Robotics Institute of Carnegie Mellon University (CMU). At present (October 2006), the toolbox only works with version 6.3 of the CMU firewire driver. Classification of the acquired image data is performed using the algorithms of the Color Machine Vision project (CMVision). This project is an initiative of the CORAL group at the Carnegie Mellon School of Computer Science. The source code of this software collection has been published under the GNU General Public License (GPL) and can be downloaded from the project’s home page at http://www-2.cs.cmu.edu/~jbruce/cmvision/. CMVision provides “a simple, robust vision system suitable for real time robotics applications” by means of “global low level colour vision at video rates without the use of special purpose hardware”.

“FireWire Vision Tools” is distributed as Free Software under the terms of the GNU General Public License Agreement (GPL). This gives a user the right to copy, re-distribute and modify the source code of the toolbox to suit any of their needs.

Frank Wornle (frank.wornle@adelaide.edu.au)

The University of Adelaide

School of Mechanical Engineering

March, 2006