ISBN 978-3-8439-1823-7

978-3-8439-1823-7, Reihe Informatik

Christian Dahmen
Robust Object Tracking for Micro- and Nanorobotics

148 Seiten, Dissertation Carl von Ossietzky Universität Oldenburg (2013), Softcover, A5

Zusammenfassung / Abstract

Image processing, specifically object tracking, is a very important technique used for the automation of micro- and nanohandling setups. Due to the lack of other sensors which deliver position information about tools and workpieces on this scale, position determination, recognition and tracking algorithms used with image data have to be used. Also, only image based approaches can cover the objects positions in relation to other objects with suffcient reliability, due to the strong influence of various effects like thermal drift, play, and the corresponding uncertainty in the relative positions and the relations between data from other peripheral sensors and the true position of the objects.

In this work, new algorithms are presented and evaluated which are robust against scanning electron microscope (SEM) movement distortions while retaining most of the advantages of previously used algorithms. The algorithms are experimentally validated and compared against the state of the art.

Also, new approaches are presented which extend two-dimensional tracking algorithms to allow extraction of three-dimensional position data from images. This includes approaches based on depth from stereo with stereo SEM imaging, and on depth from defocus. These algorithms are evaluated in experiments.

For the MRI based control of ferromagnetic objects, a new algorithm based on template stack matching is presented and evaluated in experiments. All these algorithms have been found working and stable. They can be applied in the automation of handling and manipulation setups.