Spatial localization of modelled objects of revolution in monocular perspective vision

  • M. Dhome
  • J. T. Lapreste
  • G. Rives
  • M. Richetin
Shape Description
Part of the Lecture Notes in Computer Science book series (LNCS, volume 427)


It has been demonstrated here how some geometrical features extracted from the brightness image of an object of revolution and coming from the perspective projection of lines or points situated on the surface of the object, can be used to find the spatial attitude of the object in the viewer coordinate system.

The various methods presented in this paper which involve different kinds of geometrical features are complementary since they could not be all present for some aspect of the object. For example, if the angle between the viewing direction and the axis of revolution is too small then elliptical contours will surely be seen meanwhile the projection of zero-curvature points of the scaling function will probably be hidden.

Before the design of an automatic system for localization of objects of revolution, the problem of the matching of similar contour points (zero-curvature or angular ones in the present case) has to be solved efficiently. So it is necessary to built a robust procedure to find the projection of the axis of revolution of these objects.

At last, it must be noted that for the three methods the localization problem is much simpler if at first the original viewer coordinate system is changed for a more judicious one. In fact with this new reference system, the direction of view always goes through the axis of revolution of the objects. It is worth noting that human vision proceeds in this way since the cone angle for accurate vision is rather small and consequently since the look is approximatively centered on the objects to be recognized or localized.


Object Model Perspective Projection Brightness Image Contour Point Angular Point 
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.

6 Bibliography

  1. [DHO-89]
    DHOME M., RICHETIN M, LAPRESTE J.T. & RIVES G., Determination of the Attitude of 3D Objects from a Single Perspective View, IEEE Trans. on PAMI, vol 11, no 12, pp 1265–1278, December 1989.Google Scholar
  2. [KOE-84]
    KOENDERINK J.J., What Does The Occluding Contours Tell Us About Solid Shape?, Perception, 13, pp321–331, 1984.Google Scholar
  3. [LOW-85]
    LOWE D.G. Perceptual Organization and Visual Recognition, Kluwer, Boston, 1985.Google Scholar
  4. [PON-89a]
    PONCE J., CHELLBERG D. & MANN W., Invariant Properties of Straight Homogeneous Generalized Cylinders and Their Contours, IEEE Trans. on PAMI, vol 11, no9, pp951–966, September 1989.Google Scholar
  5. [PON-89b]
    PONCE J. & KRIEGMAN D.J., On Recognizing and Positioning Curved 3D Objects from Image Contours, Proc. of IEEE Workshop on Interpretation of 3D Scenes, pp61–67, Austin, Tx, November 1989.Google Scholar
  6. [RIC-89]
    RICHETIN M., DHOME M. & LAPRESTE J.T., Inverse Perspective Transform from Zero-Curvature Curve Points. Application to the Localization of Some Generalized Cylinders, Proc. of IEEE CVPR '89, pp 517–522, San Diego, Ca, June 1989.Google Scholar
  7. [SHI-87]
    SHIRAI Y, Three Dimensional Computer Vision, Springer-Verlag, Berlin, 1987.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1990

Authors and Affiliations

  • M. Dhome
    • 1
  • J. T. Lapreste
    • 1
  • G. Rives
    • 1
  • M. Richetin
    • 1
  1. 1.Electronics Laboratory, UA 830 of the CNRSBlaise Pascal University of Clermont-FerrandAubiere CedexFrance

Personalised recommendations