Temporal Dynamical Interactions between Multiple Layers of Local Image Features for Event Detection in Video Sequences

  • Daniel Kottow
  • Mario Köppen
  • Javier Ruiz-del-Solar
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2749)

Abstract

In this paper an approach for storing and employing local image features in video processing is presented. The approach is based on the usage of memory cells representing local image features and (non-fixed) spatial positions, which are organized in memory layers. By assigning frame-based recall function and learning procedure to the cells, the memory layers establish a content-based auto-associative memory. Thus, they can be applied to solve several event detection tasks, as it is exemplified by dynamic background supression in a traffic scene, and counting of persons halting before a shopping window in an indoor scene. The case studies suggest that information gathered from the cells (like cell history based scoring values) can be used in various manners for video processing tasks circumventing the need for object segmentation and tracking, typical in many conventional background-differencing methods.

References

  1. 1.
    Institut für Algorithmen und Kognitive Systeme Kognitive Systeme-Image Sequence Server. http://i21www.ira.uka.de/image_sequences.Google Scholar
  2. 2.
    Third IEEE International Workshop on Performance Evaluation of Tracking and Surveillance. http://pets2002.visualsurveillance.org.Google Scholar
  3. 3.
    S. Richetto J.H. Piater and J.L. Crowley. Event-based activity analysis in live video using a generic object tracker. In Third IEEE Int. Workshop on Performance Evaluation of Tracking and Surveillance, 2002.Google Scholar
  4. 4.
    B.W. Weide J.L. Bentley and A.C. Yao. Optimal expected-time algorithms for closest point problems. ACM Transactions on Mathematical Software, 6(4):563–580, 1980.MATHCrossRefMathSciNetGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2003

Authors and Affiliations

  • Daniel Kottow
    • 1
    • 2
  • Mario Köppen
    • 1
    • 2
  • Javier Ruiz-del-Solar
    • 1
    • 2
  1. 1.Dept. Pattern RecognitionFraunhofer IPKBerlinGermany
  2. 2.DIEU. de ChileSantiagoChile

Personalised recommendations