Near Real-Time Pattern Recognition in a Special Purpose Computer with Parallel Architecture

  • Matthias F. Carlsohn
Conference paper
Part of the NATO ASI Series book series (NATO ASI F, volume 127)


In computer vision the term real-time possesses some uncertainty, because of the demands defined by standard camera sensors and the requirements of the processes under inspection sometimes differ by orders. An example shows the necessary system complexity. In image pattern recognition, the segmentation of interesting image objects from their image background and the characterization of the object properties by their describing features are the pre-requisites for an object classification. Both process steps are usually of great computational complexity and time consumption, respectively. Consequently, a processing in video real-time is only possible by a supporting computer architecture.


Object Classification Parallel Architecture Object Candidate Real Time Computing Graceful Degradation 
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.
    C. Anderer, U. Thönnessen, M.F. Carlsohn, A. Klonz, Ein Bildsegmentierer für die echtzeitnahe Verarbeitung, 11. DAGM Symposium Mustererkennung, 2.4. Oktober 1989, Hamburg, Informatik-Fachbericht 219, K. Burkhardt, K.H. Höhne, B. Neumann (Eds.), Springer-Verlag, Berlin, Heidelberg 1989, 380–384.Google Scholar
  2. 2.
    U. Thönnessen, D. Ernst, H. Gro, Entwicklung einer segmentbasierten Beschreibung von Ereignissen in Bildfolgen, 13. DAGM Symposium Mustererkennung, 9.-11. Oktober 1991, Miinchen, Informatik-Fachbericht 290, B. Radig (Ed.), Springer-Verlag, Berlin, Heidelberg 1991, 499–506.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1994

Authors and Affiliations

  • Matthias F. Carlsohn
    • 1
  1. 1.BremenGermany

Personalised recommendations