Lecture Notes in Computer Science 5604 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 Alfred Kobsa University of California, Irvine, CA, 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 Gerhard Weikum Max-Planck Institute of Computer Science, Saarbruecken, Germany
Daniel Cremers Bodo Rosenhahn Alan L. Yuille Frank R. Schmidt (Eds.) Statistical and Geometrical Approaches to Visual Motion Analysis International Dagstuhl Seminar Dagstuhl Castle, Germany, July 13-18, 2008 Revised Papers 13
Volume Editors Daniel Cremers Universität Bonn, Institut für Informatik III Römerstraße 164, 53117 Bonn, Germany E-mail: dcremers@cs.uni-bonn.de Bodo Rosenhahn Leibniz Universität Hannover, Institut für Informationsverarbeitung Appelstraße 9A, 30167 Hannover, Germany E-mail: rosenhahn@tnt.uni-hannover.de Alan L. Yuille University of California, Department of Statistics and Psychology 8967 Math Sciences Building, Los Angeles, CA 90095-1554, USA E-mail: yuille@stat.ucla.edu Frank R. Schmidt Universität Bonn, Institut für Informatik III Römerstraße 164, 53117 Bonn, Germany E-mail: schmidtf@cs.uni-bonn.de Library of Congress Control Number: 2009930038 CR Subject Classification (1998): I.4.8, I.4, I.2.8-10, I.5, I.3.5, F.2.2 LNCS Sublibrary: SL 6 Image Processing, Computer Vision, Pattern Recognition, and Graphics ISSN 0302-9743 ISBN-10 3-642-03060-2 Springer Berlin Heidelberg New York ISBN-13 978-3-642-03060-4 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.com Springer-Verlag Berlin Heidelberg 2009 Printed in Germany Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India Printed on acid-free paper SPIN: 12708987 06/3180 543210
Preface Motion analysis is central to both human and machine vision. It involves the interpretation of image data over time and is crucial for a range of motion tasks such as obstacle detection, depth estimation, video analysis, scene interpretation, video compression and other applications. Motion analysis is unsolved because it requires modeling of the complicated relationships between the observed image data and the motion of objects and motion patterns (e.g., falling rain) in the visual scene. The Dagstuhl Seminar 08291 on Statistical and Geometrical Approaches to Visual Motion Analysis was held during July 13 18, 2008 at the International Conference and Research Center (IBFI), Schloss Dagstuhl, near Wadern in Germany. The workshop focused on critical aspects of motion analysis, including motion segmentation, the modeling of motion patterns and the different techniques used. These techniques include variational approaches, level set methods, probabilistic models, graph cut approaches, factorization techniques, and neural networks. All these techniques can be subsumed within statistical and geometrical frameworks. We further involved experts in the study of human and primate vision. Primate visual systems are extremely sophisticated at processing motion, thus there is much to be learnt from studying them. In particular, we discussed how to relate the computational models of primate visual systems to those developed for machine vision. In total, 15 papers were accepted for these proceedings after the workshop. We were careful to ensure a high standard of quality for the accepted papers. All submissions were double-blind reviewed by at least two experts. The accepted papers reflect the state of the art in the field and cover various topics related to motion analysis. The papers in this volume are classified into four categories based on the topics optic flow and extensions, human motion modeling, biological and statistical approaches and alternative approaches to motion analysis. We would like to thank the team at castle Dagstuhl for the professional support before, after and during the seminar. We are grateful to the participants of the Dagstuhl workshop for their active discussions and commitment during the seminar and for the remarkable efforts and the quality of timely delivered reviews. Apart from all the authors, we would like to also thank Christoph Garbe, Timo Kohlberger and Thomas Schoenemann for providing additional reviews. The organization of this event would not have been possible without the effort and the enthusiasm of several people, and we thank all who contributed. April 2009 Daniel Cremers Bodo Rosenhahn Alan Yuille Frank R. Schmidt
Table of Contents Optical Flow and Extensions Discrete-Continuous Optimization for Optical Flow Estimation... 1 Stefan Roth, Victor Lempitsky, and Carsten Rother An Improved Algorithm for TV-L 1 Optical Flow... 23 Andreas Wedel, Thomas Pock, Christopher Zach, Horst Bischof, and Daniel Cremers An Evaluation Approach for Scene Flow with Decoupled Motion and Position... 46 Andreas Wedel, Tobi Vaudrey, Annemarie Meissner, Clemens Rabe, Thomas Brox, Uwe Franke, and Daniel Cremers An Affine Optical Flow Model for Dynamic Surface Reconstruction... 70 Tobias Schuchert and Hanno Scharr Deinterlacing with Motion-Compensated Anisotropic Diffusion... 91 Matthias Ghodstinat, Andrés Bruhn, and Joachim Weickert Human Motion Modeling Real-Time Synthesis of Body Movements Based on Learned Primitives... 107 Martin A. Giese, Albert Mukovskiy, Aee-Ni Park, Lars Omlor, and Jean-Jacques E. Slotine 2D Human Pose Estimation in TV Shows... 128 Vittorio Ferrari, Manuel Marín-Jiménez, and Andrew Zisserman Recognition and Synthesis of Human Movements by Parametric HMMs... 148 Dennis Herzog and Volker Krüger Recognizing Human Actions by Their Pose... 169 Christian Thurau and Václav Hlaváč Biological and Statistical Approaches View-Based Approaches to Spatial Representation in Human Vision... 193 Andrew Glennerster, Miles E. Hansard, and Andrew W. Fitzgibbon
VIII Table of Contents Combination of Geometrical and Statistical Methods for Visual Navigation of Autonomous Robots... 209 Naoya Ohnishi and Atsushi Imiya Motion Integration Using Competitive Priors... 235 Shuang Wu, Hongjing Lu, Alan Lee, and Alan Yuille Alternative Approaches to Motion Analysis Derivation of Motion Characteristics Using Affine Shape Adaptation for Moving Blobs... 259 Jorge Sanchez, Reinhard Klette, and Eduardo Destefanis Comparison of Point and Line Features and Their Combination for Rigid Body Motion Estimation... 280 Florian Pilz, Nicolas Pugeault, and Norbert Krüger The Conformal Monogenic Signal of Image Sequences... 305 Lennart Wietzke, Gerald Sommer, Oliver Fleischmann, and Christian Schmaltz Author Index... 323