	US 7,457,436 B2
	Real-time crowd density estimation from video
Nikos Paragios, Cranbury, N.J. (US); Visvanathan Ramesh, Plainsboro, N.J. (US); Bjoern Stenger, Cambridge (United Kingdom); and Frans Coetzee, Princeton, N.J. (US)
Assigned to Siemens Corporate Research, Inc., Princeton, N.J. (US)
Filed on Oct. 10, 2006, as Appl. No. 11/545,236.
Application 11/545236 is a continuation of application No. 09/944317, filed on Aug. 31, 2001, granted, now 7,139,409.
Claims priority of provisional application 60/230264, filed on Sep. 06, 2000.
Prior Publication US 2007/0031005 A1, Feb. 08, 2007
Int. Cl. G06K 9/00 (2006.01); H04N 5/225 (2006.01)

U.S. Cl. 382—103 [382/160; 348/169]

15 Claims

10. A method of measuring congestion in an optical system, comprising:

estimating a background reference frame for representing a background;

estimating geometric parameters for representing a scale variation of objects in a given frame;

obtaining a change detection map for distinguishing the background from the objects in the given frame;

combining the change detection map with the geometric parameters to determine a measure of congestion of the given frame;

initializing each region of the image with a single node and a local model;

evaluating confidence limits of the local model; and

evaluating the local model to determine a multi-modality, wherein if a multi-modality is detected the local model is split into multiple nodes.