Multi-agent Based Optic Flow

Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 193)


In this article, the authors present a novel algorithm for computing optic flow using a multi-agent based feature point tracking method. In this multi-agent based optic flow method, feature points which are invariant to scale, orientation and illumination changes are extracted and tracked in parallel using independent agents. Each agent is run by a separate light-weight thread which can be implemented using parallel processes on a multicore processor. The agents use a Kalman filter to predict the frame to frame position of the feature points in the image, producing position and velocity data for each feature point, which can then be used to perform optic flow, while simultaneously producing feature descriptors that can be used for object recognition and stereopsis. We show that in a parallel implementation, this algorithm provides significant performance advantages over other feature point tracking object recognition methods. It therefore may provide a plausible basis for a unified computer vision architecture including optic flow, object recognition, and stereopsis.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Lowe, D.G.: Distinctive Image Features from Scale-Invariant Keypoints. Journal of Computer Vision (2004)Google Scholar
  2. 2.
    Lucas, B.D., Kanade, T.: An iterative image registration technique with an application to stereo vision. In: Imaging Understanding Workshop, pp. 121–130 (1981)Google Scholar
  3. 3.
    Horn, B.K.P., Schunck, B.G.: Determining optical flow. AI 17, 185–204 (1981)Google Scholar
  4. 4.
    Subramaniam, S., Biederman, I., Kalocsai, P., Madigan, S.R.: Accurate identification, but chance forced-choice recognition for rsvp pictures.In: Annual Meeting of the Association for Research in Vision and Ophthalmology, Ft. Lauderdale, FL (1995)Google Scholar
  5. 5.
    DiGirolamo, G., Kanwisher, N.: Accessing stored representations begins within 155 ms in object recognition. In: 36th Annual Meeting of the Psychonomics Society, Los Angeles (1995)Google Scholar
  6. 6.
    Harris, C., Stephens, M.: A combined corner and edge detector. In: Fourth Alvey Vision Conference, Manchester, UK, pp. 147–151 (1988)Google Scholar
  7. 7.
    Mikolajczyk, K., Schmid, C.: An Affine Invariant Interest Point Detector. In: Heyden, A., Sparr, G., Nielsen, M., Johansen, P. (eds.) ECCV 2002, Part I. LNCS, vol. 2350, pp. 128–142. Springer, Heidelberg (2002)CrossRefGoogle Scholar
  8. 8.
    Beis, J., Lowe, D.G.: Shape indexing using approximate nearest-neighbour search in highdimensional spaces. In: Conference on Computer Vision and Pattern Recognition, Puerto Rico, pp. 1000–1006 (1997)Google Scholar
  9. 9.
    Beis, J.S.: Indexing without invariants in model based object recognition. PhD thesis, University of British Columbia (1997)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  1. 1.Drexel UniversityPhiladelphiaUSA
  2. 2.Iguana Robotics IncTucsonUSA

Personalised recommendations