AUTOMATED MEASUREMENT OF CROWD DENSITY AND MOTION USING IMAGE PROCESSING

Earlier methods of measuring pedestrian movement by manual or automated means are reviewed and new techniques described which are independent of the shape, size and velocity of individual pedestrians. Two pixel-based methods for estimating crowd density are described: a) thinned edges and b) background removal. The use of head movements as a measure of pedestrian flow is described. Methods of estimating motion are considered. A method is described based on grey-level correlation of small neighbourhoods in two consecutive frames. When the data is aggregated, good correlation with actual pedestrian movements is found in the results from the background removal and frame difference methods. For the covering abstract see IRRD 865401.

Language

  • English

Media Info

  • Features: References;
  • Pagination: p. 127-32

Subject/Index Terms

Filing Info

  • Accession Number: 00668591
  • Record Type: Publication
  • Source Agency: Transport Research Laboratory
  • Files: ITRD, ATRI
  • Created Date: Nov 16 1994 12:00AM