Beta 1


Title Observation of Airport Queue Sizes using Surveillance Cameras
Author Jensen, Martin Andreas Falk
Supervisor Larsen, Rasmus (Image Analysis and Computer Graphics, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Master's thesis
Year 2009
Abstract This thesis deals with the development and evaluation of a system that with the use of image analysis and mathematical modeling, can measure the size of queues from image sequences recorded with surveillance cameras. The system is developed with the aim of solving the image analysis problem defined by a request from SAS Scandinavian Airlines. SAS requested a method to automatically measure queue sizes in longer periods of time in Copenhagen Airport from a large database of image sequences. The problem is solved based on the assumption that queues can be identified as foreground objects in the image sequence. The foreground segmentation of the images is done with a modified version of the Mixture of Gaussian’s method by Stauffer and Grimson [13] that can handle most of the dynamic changes of the queue background in an indoor environment. The extent of the queue on the floor plane is subsequently estimated by assuming that the height of the queue is relatively constant and applying knowledge of the image formation in the surveillance camera. Finally the relative size of the queue is predicted by a linear regression model that predicts the queue size as a linear combination of the area of the queue extent in different parts of the floor plane. The developed system has been evaluated on two queue locations and is capable of categorizing the queue size in long periods of time into four groups with an average classification rate of about 85%. The classification rate can be improved if the number of groups are reduced to three. The system can be applied on other queue locations with few preparations. It is concluded that the developed system can produce reliable estimates of the queue size over longer periods of time.
Note Thesis not publicly available
Imprint Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Series IMM-M.Sc.-2009-70
Admin Creation date: 2009-12-09    Update date: 2009-12-09    Source: dtu    ID: 253642    Original MXD