Motion Estimation Revisited: An Estimation-Theoretic Approach

The talk and the associated paper analyze some previously unexplored aspects of motion estimation that are fundamental both for discrete block matching as well as for differential ’optical flow’ approaches `a la Lucas-Kanade. It aims at providing a complete estimation-theoretic approach that makes the assumptions about noisy observations of samples from a continuous signal of a certain class explicit. It turns out that motion estimation is a combination of simultaneously estimating the true underlying continuous signal and optimizing the displacement between two hypothetical copies of this unknown signal. Practical schemes such as the current variants of Lucas-Kanade are just approximations to the fundamental estimation problem identified in the present paper. Derivatives appear as derivatives to the continuous signal representation kernels, not as ad hoc discrete derivative masks. The formulation via an explicit signal space defined by kernels is a precondition for analyzing e.g. the convergence range of iterative displacement estimation procedures, and for systematically choosing preconditioning filters.

Speaker Details

Rudolf Mester obtained a Dipl.-Ing. (1983) and Dr.-Ing. degree (1988) in electrical engineering from RWTH Aachen, Germany. In 1990, Dr. Mester established the computer vision and image interpretation group at the Research Institute of Robert Bosch GmbH, Hildesheim, which puts special emphasis on applying computer vision to surveillance, security and driver assistance (ADAS). In October 1995, Dr. Mester was appointed professor at Goethe University, Frankfurt am Main (Physics dept. 1995, Computer Science 2005). Currently, his research interests are focused on statistical signal and image processing methods, the construction of robust and reliable vision algorithms and flexible vision systems as well as the theoretical foundations for ‘seeing machines’. Dr. Mester was a co-initiator and from 2008 – 2011 the coordinator of the U Frankfurt contribution to the Bernstein Focus on Neurotechnology Frankfurt adressing vision in man and machine. From 4/2011 to 3/2012 and from 10/2012 – present, Rudolf Mester is also a visiting professor in the Computer Vision Laboratory, E.E. Dept. (ISY) of Linköping University, Sweden.

Date:
Speakers:
Rudolf Mester
Affiliation:
J.W. Goethe-University
    • Portrait of Jeff Running

      Jeff Running