Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/2642
Title: Human motion detection and tracking in videos
Authors: Guo, Jing
Keywords: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
Issue Date: 2007
Source: Guo, J. (2007).Human motion detection and tracking in videos. Master’s thesis, Nanyang Technological University, Singapore.
Abstract: Human motion detection and tracking are very important research areas in vision based video processing. This thesis presents our efforts to understand and improve human motion detection and tracking techniques. We examined the existing motion detection approaches and proposed our own methods. The first is entropy based motion detection, where the entropy of accumulated difference image is used to detect motion regions. Another simple yet effective motion detection method calculates the optimum threshold for every difference image. Single camera tracking follows motion detection and a mean shift tracking algorithm is firstly implemented. Its advantages and limitations are analyzed and possible improvements are given. We then proposed a single camera tracking method by corresponding detected motion regions. Experimental results demonstrate the effectiveness of our algorithm. A motion detection method based affine tracking method is also presented. Finally, some multi-camera tracking problems are addressed. A homography based method is used for cross camera correspondence. Given multiple cameras, more information is available for the same person. Some preliminary results on selecting the front view of a walking person is shown. Besides, a literature survey on the related areas is also included in the thesis.
URI: https://hdl.handle.net/10356/2642
DOI: 10.32657/10356/2642
Rights: Nanyang Technological University
Fulltext Permission: open
Fulltext Availability: With Fulltext
Appears in Collections:SCSE Theses

Files in This Item:
File Description SizeFormat 
SCE-THESES_68.pdf2.43 MBAdobe PDFThumbnail
View/Open

Google ScholarTM

Check

Altmetric


Plumx

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.