Please use this identifier to cite or link to this item:
Title: Multiple object tracking by stochastic method
Authors: Li, Jiang
Keywords: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Control engineering
DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
Issue Date: 2006
Source: Li, J. (2006). Multiple object tracking by stochastic method. Doctoral thesis, Nanyang Technological University, Singapore.
Abstract: Visual tracking and surveillance has become an active research area of computer vision due to the demand from the public for improved security and safety. The research work of visual tracking and surveillance involves many technical issues, such as motion segmentation, object representation, object tracking and behavior understanding. Among these issues, object representation and tracking are specially important. Being able to maintain tracking of objects in video sequences is not only useful by itself but also a crucial step to higher level video interpretation. The problems are made difficult due to non-stationary environment, persistent and temporary occlusion of multiple interacting objects and low image resolution in cases of distant viewing, which occur frequently in real applications. A stochastic transductive adaptation method is proposed in this thesis to address the problem of non-stationary object tracking in a complex environment. The proposed stochastic transductive adaptation algorithm combines stochastic transductive learning with a locally exploring particle filter. This adaptive tracker can efficiently and successfully handle on-rigid objects under different appearance changes by its stochastic transductive learning ability. Objects can be tracked well despite severe occlusion or clutter.
DOI: 10.32657/10356/3433
Rights: Nanyang Technological University
Fulltext Permission: open
Fulltext Availability: With Fulltext
Appears in Collections:EEE Theses

Files in This Item:
File Description SizeFormat 
EEE-THESES_1296.pdf4.15 MBAdobe PDFThumbnail

Page view(s) 50

Updated on May 12, 2021

Download(s) 5

Updated on May 12, 2021

Google ScholarTM




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