Go to  Advanced Search

Background subtraction with a pan/tilt camera

Show full item record

Files in this item

Files Size Format Description   View
ubc_2011_spring_tsinko_egor.pdf 14.57Mb Adobe Portable Document Format   View/Open
Title: Background subtraction with a pan/tilt camera
Author: Tsinko, Egor
Degree Master of Science - MSc
Program Computer Science
Copyright Date: 2010
Publicly Available in cIRcle 2010-12-13
Abstract: Background subtraction is a common approach in computer vision to detect moving objects in video sequences taken with a static camera. Some background subtraction algorithms have been extended to support a pan/tilt camera but they are required to provide a background mosaic for motion estimation purposes. This thesis describes a system that avoids this requirement by separating the motion estimation system and background subtraction system into two separate modules. The first module performs motion compensation by employing a feature based image registration method and the RANSAC algorithm. The second module detects moving objects in rectified image frames using a background subtraction algorithm designed for a static camera. Three background subtraction algorithms were extended in the course of this project: mixture of Gaussians, non-parametric kernel density estimation and codebook. This thesis demonstrates the usefulness of separating of motion estimation from the background subtraction system as it allows us to extend any background subtraction algorithm designed for a static camera to support a pan/tilt camera. The detection results are presented for both indoor and outdoor video sequences taken with a pan/tilt camera.
URI: http://hdl.handle.net/2429/30386
Scholarly Level: Graduate

This item appears in the following Collection(s)

Show full item record

All items in cIRcle are protected by copyright, with all rights reserved.

UBC Library
1961 East Mall
Vancouver, B.C.
Canada V6T 1Z1
Tel: 604-822-6375
Fax: 604-822-3893