How to achieve perception of real-world scenes: A case study
As a case study for a complex real-world perception task we investigate a video-based system for traffic monitoring. We show how this application can be handled economically using a robust low-level background update routine, cheap, but carefully selected physical and geometric models of the relevant objects and a cooperation between the various processing subtasks (detection, tracking, classification).
Unable to display preview. Download preview PDF.
- /1/.Wolf, J.K., Viterbi, A.M., Dixon, G.S.: Trellis decoding and application to multi-target tracking. Communication Networking in Dense Environments. Los Angeles, CA, USA, 14–15 Jan 1988, Proc. SPIE–Int. Opt. Eng. (USA), vol. 876.Google Scholar
- /2/.Karmann, K.-P., v. Brandt, A.: Moving Object Recognition Using an Adaptive Background Memory. Proc. of the 6th Scandinavian Conference on Image Analysis, Oulu, Finland, June 19–22, 1989, pp. 1051–1058.Google Scholar
- /3/.Karmann, K.-P., v. Brandt, A., Gerl, R.: Moving Object Segmentation Based on Adaptive Reference Images. Proc. EUSIPCO-90, Barcelona, 1990, pp. 951–954Google Scholar