Dissertation/Thesis Abstract

A particle filter methodology for salient object detection in videos
by Soma, Venkata Manikrishna, M.S., California State University, Long Beach, 2016, 31; 10142985
Abstract (Summary)

Object tracking in a video consists of continuous non-stationary images, which change with time. In the computer research field, tracking a moving object is a complicated issue. Many existing algorithms are only capable in controlled and predefined conditions. This project proposes improvements in color based tracking to track a moving object. Particle filtering is successful for non-Gaussian and non-linear problems. The object state has been taken as the object position, speed, size, object size scale, and the object’s appearance. The target model’s update condition and adaptive likelihood are calculated to ensure the proper object tracking. The state space model is used for particle movement. “State” refers to the position of a particle. The particle likelihood is calculated based on the closeness between the particle pixel position and the target particle color. The experimental results prove that the object is successfully tracked by the particle filter.

Indexing (document details)
Advisor: Yeh, Hen-Guel
Commitee: Ary, James, Wang, Fei
School: California State University, Long Beach
Department: Electrical Engineering
School Location: United States -- California
Source: MAI 55/06M(E), Masters Abstracts International
Subjects: Electrical engineering
Publication Number: 10142985
ISBN: 978-1-339-98239-7
Copyright © 2020 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy