Motion segmentation and depth ordering based on morphological segmentation

  • Lothar Bergen
  • Fernand Meyer
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1407)


In this paper the motion segmentation and depth ordering problem for monocular image sequences with and without camera motion is addressed. We show how a new multiscale morphological segmentation technique, based on the watershed, can produce a superset of the motion boundaries. Regions with similar motion then have to be merged. The difficulties of motion estimation at object boundaries with occlusion are analyzed and a solution combining segmentation and robust estimation is presented. Region merging is then performed using the obtained motion parameters. We then present a new technique for the depth ordering of the resulting image partition. We show how the modelling error on either side of the motion boundary can be used to indicate the occlusion relationship of the objects. The algorithm is then applied to several synthetic and natural image sequences. The results demonstrate that the technique is robust and that the depth ordering requires only minimal motion to perform correctly. This is due to the fact that, unlike existing techniques for depth ordering, the motion between two frames only has to be analyzed. We then point out possible improvements and indicate how temporal integration of the information can further increase stability.


Motion Estimation Motion Parameter Object Boundary Motion Boundary Motion Measurement 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    J. Cichosz and F. Meyer. Morphological multiscale image segmentation. In Workshop on Image Analysis for Multimedia Interactive Services (WIAMIS'97), pages 161–166, Louvain-la-Neuve (Belgium), June 1997.Google Scholar
  2. 2.
    Trevor Darrell and David Fleet. Second-order method for occlusion relationships in motion layers. Technical Report 314, MIT Media Lab Vismod, 1995.Google Scholar
  3. 3.
    E. Decencière Ferrandière, C. de Fouquet, and F. Meyer. Applications of kriging to image sequence coding. Accepted for publication in Signal Processing: Image Communication, 1997.Google Scholar
  4. 4.
    B. K. P. Horn and B. G. Schunck. Determining optical flow. Artificial Intelligence, 17:185–203, 1981.CrossRefGoogle Scholar
  5. 5.
    Peter Meer, Doron Mintz, Dong Yoon Kim, and Azriel Rosenfeld. Robust regression methods for computer vision: A review. International Journal of Computer Vision, 6(1):59–70, April 1991.CrossRefGoogle Scholar
  6. 6.
    K. M. Mutch and W. B. Thompson. Analysis of accretion and deletion at boundaries in dynamic scenes. IEEE Transactions on Pattern Analysis and Machine Intelligence, 7:133–138, 1985.CrossRefGoogle Scholar
  7. 7.
    W. B. Thompson, K. M. Mutch, and V. A. Berzins. Dynamic occlusion analysis in optical flow fields. IEEE Transactions on Pattern Analysis and Machine Intelligence, 7:374–383, 1985.Google Scholar
  8. 8.
    Zhengyou Zhang. Parameter estimation techniques: A tutorial with application to conic fitting. Technical Report 2676, Institut National de Recherche en Informatique et en Automatique, Sophia-Antipolis Cedex, France, October 1995.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1998

Authors and Affiliations

  • Lothar Bergen
    • 1
  • Fernand Meyer
    • 1
  1. 1.Centre de Morphologie MathématiqueEcole des Mines de ParisFontainebleau CedexFrance

Personalised recommendations