Lecture Notes in Computer Science 3417

Transcription

1 Lecture Notes in Computer Science 3417 Commenced Publication in 1973 Founding and Former Series Editors: Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen Editorial Board David Hutchison Lancaster University, UK Takeo Kanade Carnegie Mellon University, Pittsburgh, PA, USA Josef Kittler University of Surrey, Guildford, UK Jon M. Kleinberg Cornell University, Ithaca, NY, USA Friedemann Mattern ETH Zurich, Switzerland John C. Mitchell Stanford University, CA, USA Moni Naor Weizmann Institute of Science, Rehovot, Israel Oscar Nierstrasz University of Bern, Switzerland C. Pandu Rangan Indian Institute of Technology, Madras, India Bernhard Steffen University of Dortmund, Germany Madhu Sudan Massachusetts Institute of Technology, MA, USA Demetri Terzopoulos University of California, Los Angeles, CA, USA Doug Tygar University of California, Berkeley, CA, USA Moshe Y. Vardi Rice University, Houston, TX, USA Gerhard Weikum Max-Planck Institute of Computer Science, Saarbruecken, Germany

2 Bernd Jähne Erhardt Barth Rudolf Mester (Eds.) Complex Motion First International Workshop, IWCM 2004 Günzburg, Germany, October 12-14, 2004 Revised Papers 13

3 Volume Editors Bernd Jähne Universität Heidelberg IWR und Institut für Umweltphysik Im Neuenheimer Feld 368, Heidelberg, Germany Rudolf Mester J.W. Goethe Universität Institut für Informatik, vsi Arbeitsgruppe Robert-Mayer-Str. 10, Frankfurt, Germany Erhardt Barth Universität Lübeck Institut für Neuro- und Bioinformatik Ratzeburger Allee 160, Lübeck, Germany Forschungszentrum Jülich GmbH Institut für Chemie und Dynamik der Geosphäre Institut III: Phytosphäre, Jülich, Germany Library of Congress Control Number: CR Subject Classification (1998): I.5, I.4, I.3.5, I.2.10, I.2.6, F.2.2 LNCS Sublibrary: SL 6 Image Processing, Computer Vision, Pattern Recognition, and Graphics ISSN ISBN Springer Berlin Heidelberg New York ISBN Springer Berlin Heidelberg New York This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. Springer is a part of Springer Science+Business Media springer.com Springer-Verlag Berlin Heidelberg 2007 Printed in Germany Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India Printed on acid-free paper SPIN: /

4 Preface The world we live in is a dynamic one: we explore it by moving through it, and many of the objects which we are interested in are also moving. Traffic, for instance, is an example of a domain where detecting and processing visual motion is of vital interest, both in a metaphoric as well as in a purely literal sense. Visual communication is another important example of an area of science which is dominated by the need to measure, understand, and represent visual motion in an efficient way. Visual motion is a subject of research which forces the investigator to deal with complexity; complexity in the sense of facing effects of motion in a very large diversity of forms, starting from analyzing simple motion in a changing environment (illumination, shadows,...), under adverse observation conditions, such as bad signal-to-noise ratio (low illumination, small-scale processes, low-dose x-ray, etc.), covering also multiple motions of independent objects, occlusions, and going as far as dealing with objects which are complex in themselves (articulated objects such as bodies of living beings). The spectrum of problems includes, but does not end at, objects which are not bodies at all, e.g., when analyzing fluid motion, cloud motion, and so on. Analyzing the motion of a crowd in a shopping mall or in an airport is a further example that implies the need to struggle against the problems induced by complexity. We cannot be sure that the named or similar application areas already represent the high-end of complexity in motion analysis actually, there will probably be even harder problems in the analysis of complex visual motion which we have not faced yet, or which we have not yet dared to address. Based on the observation that the current state of the art in the field of motion analysis is in a rather advanced shape already, but also taking into account that there are so many real-life problems which have not been solved yet, a group of researchers from different German research institutions decided to initiate an international workshop to attract renowned scientists and young researchers from different areas of visual motion analysis. Therefore, in October 2004, the 1st International Workshop on Complex Motion (IWCM 2004) was held at Schloss Günzburg, a beautiful mansion and scientific convention center administered by the University of Ulm (South Germany). The Steering Committee of IWCM 2004 aimed at inspiring and encouraging the members of the computer vision community to share experiences and exchange opinions on the contemporary development of the field. There were several invited talks given by renowned senior researchers who not only appreciated the historic development of research in visual motion, but demonstrated and discussed the current grand challenges in a vivid and stimulating way. This workshop was particularly devoted to advancing the repertoire of methods dealing with complex visual motion and to initiating a more intensive and

5 VI Preface hopefully continuing discussion amongst leading experts in the field. The topics of presentations were optical flow, local motion estimation for image signals affected by strong disturbances, structure from motion, multicamera flow analysis, dynamic stereo, fluid motion analysis, the estimation of multiple motions, motion tracking, and many other areas where complex visual motion patterns have to be evaluated. In fact, there were several plenary discussions which covered open issues, unsolved problems, and also different opinions in a highly constructive manner and which apparently ignited many further discussions adjacent to the official workshop programme, and presumably initiated exchange and cooperation between researchers who had not been in direct contact before. The workshop was organized by the members of the LOCOMOTOR Project (Nonlinear analysis of multidimensional signals: LOcal adaptive estimation of COmplex MOTion and ORientation patterns), which is part of the priority research program SPP 1114 Mathematical methods for time series analysis and digital image processing supported by the German Research Council (DFG). We particularly appreciate the generous support that our workshop received from the German Pattern Recognition Association (DAGM, Deutsche Arbeitsgemeinschaft für Mustererkennung) and its president, Prof. Dr. Hans Burkhardt (University of Freiburg). Without this generous support this workshop would not have been possible. We also appreciate the kind support of Springer for giving us the opportunity to distribute the scientific essence of this workshop to the computer science and engineering community via a volume in the Springer LNCS series. We hope that the present compilation of research papers presented at IWCM 2004 reflects the diversity of challenges, the prosperity of the research field, and possibly also a bit of the enjoyable atmosphere we shared at Schloss Günzburg. November 2006 Rudolf Mester Bernd Jähne Erhardt Barth

6 Organization Steering Committee Rudolf Mester Bernd Jähne Erhardt Barth Frankfurt University (Programme Chair, overall management) Heidelberg University (Workshop Proceedings) Lübeck University Research Center Jülich Sponsoring Institutions German Pattern Recognition Association (DAGM) German Research Council (DFG)

7 Table of Contents Optical Flow Estimation from Monogenic Phase... 1 Michael Felsberg Optimal Filters for Extended Optical Flow Wiener-Optimized Discrete Filters for Differential Motion Estimation Kai Krajsek and Rudolf Mester Boundary Characterization Within the Wedge-Channel Representation Ullrich Köthe Multiple Motion Estimation Using Channel Matrices Per-Erik Forssén and Hagen Spies Divide-and-Conquer Strategies for Estimating Multiple Transparent Motions Cicero Mota, Ingo Stuke, Til Aach, and Erhardt Barth Towards a Multi-camera Generalization of Brightness Constancy Complex Motion in Environmental Physics and Live Sciences Bernd Jähne Bayesian Approaches to Motion-Based Image and Video Segmentation Daniel Cremers On Variational Methods for Fluid Flow Estimation Paul Ruhnau, Jing Yuan, and Christoph Schnörr Motion Based Estimation and Representation of 3D Surfaces and Boundaries Klas Nordberg and Fredrik Vikstén A Probabilistic Formulation of Image Registration Christoph Strecha, Rik Fransens, and Luc Van Gool Myocardial Motion and Strain Rate Analysis from Ultrasound Sequences Michael Sühling, Muthuvel Arigovindan, Christian Jansen, Patrick Hunziker, and Michael Unser

8 X Table of Contents Determining the Translational Speed of a Camera from Time-Varying Optical Flow Anton van den Hengel, Wojciech Chojnacki, and Michael J. Brooks A Robust Approach for Ego-Motion Estimation Using a Mobile Stereo Platform Hernán Badino Robust Monocular Detection of Independent Motion by a Moving Observer Felix Woelk and Reinhard Koch Tracking Complex Objects Using Graphical Object Models Leonid Sigal, Ying Zhu, Dorin Comaniciu, and Michael Black Author Index