Visual Form pp 313-321 | Cite as

Depth Data Segmentation Using Robust Estimation in a Hierarchical Data Structure

  • G. A. Jones
  • J. Princen
  • J. Illingworth
  • J. Kittler


In this paper a novel segmentation algorithm which fits parametric surfaces to depth data is presented. The method incorporates Robust Statistical Estimation techniques [3] together with a hierachical data structure. The Robust Statistical techniques are important as they provide both outlier rejection and some tolerance to deviations from modelled th data distribution. The outlier rejection effectively corresponds to constraining the points which enter the fit to lie within a 3D template region. Deviations from commonly assummed model distributions (such as Gaussian noise processes) can occur from many effects including inclusion of two or more physical surfaces in a fitting window. The use of a hierarchy is important as the smallest image windows are likely to contain simple scene structures. Therefore the surface parameters derived from this lowest level are likely to be accurate representations of the window contents. These estimates can then be used to provide strong constraints which allow reliable and increasingly accurate surface estimates to be calculated at higher levels of the hierarchy. In this paper we demonstrate these principles by developing an algorithm which fits planar surfaces to depth data. Results are shown on both synthetic and real depth data images.


Depth Data Influence Function Hough Transform Outlier Rejection Sibling Node 
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.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    P.J. Besl, J.B. Birch, and L.T. Watson. Robust window operators. In Proceedings of IEEE 2 nd Int. Conf. Computer Vision, pp 591-600, Florida, December 1988.Google Scholar
  2. [2]
    P.J. Besl and R.C. Jain. Segmentation through variable-order surface fitting. IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(2):167–192, March 1988.CrossRefGoogle Scholar
  3. [3]
    F. Hampel, E. Ronchetti, P. Rousseeuw, and W. Stahel. Robust Statistics: The approach based on Influence Functions. John Wiley and Sons, 1986.Google Scholar
  4. [4]
    G.A. Jones, J. Illingworth, and J. Kittler. Robust local window processing of range images. In Proceedings 7th Scand. Conf. Image Analysis, Denmark, August 1991.Google Scholar
  5. [5]
    A. Leonardis, A. Gupta, and R. Bajcsy. Segmentation as the search for the best description of the image in terms of primitives. In Proceedings of 3 rd IEEE International Conference on Computer Vision, pp 121-125, Osaka, Japan, December 1990.Google Scholar
  6. [6]
    Y. Muller and R. Mohr. Planes and quadric detection using the Hough transform. In Proceedings of IEEE Int. Conf. Pattern Recognition, pp 1101-1103, 1984.Google Scholar
  7. [7]
    T. Nagata and B. Zha. Determining orientation, location and size of primitive surfaces by a modified Hough transformation technique. Patt.Recog, 21(5):481–491, 1988.CrossRefGoogle Scholar
  8. [8]
    J. Princen. Hough Transform Methods for Curve Detection and Parameter Estimation. PhD thesis, University of Surrey, June 1990.Google Scholar
  9. [9]
    J. Princen, J. Illingworth, and J. Kittler. A hierarchical approach to line extraction. In Computer Vision, Graphics and Image Processing, Vol 52, pp 57–77, Sept 1990.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 1992

Authors and Affiliations

  • G. A. Jones
    • 1
  • J. Princen
    • 1
  • J. Illingworth
    • 1
  • J. Kittler
    • 1
  1. 1.Department of Electronic and Electrical EngineeringUniversity of SurreyGuildfordUK

Personalised recommendations