The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Optical flow determination and motion analysis
Author:	Wang, Qian
View Online:	njit-etd1993-159 (x, 85 pages ~ 2.6 MB pdf)
Department:	Department of Electrical and Computer Engineering
Degree:	Master of Science
Program:	Electrical Engineering
Document Type:	Thesis
Advisory Committee:	Shi, Yun Q. (Committee chair) Hou, Edwin (Committee member) Kuo, Marshall Chuan Yung (Committee member)
Date:	1993-01
Keywords:	Image processing -- Digital techniques Motion -- Analysis -- Automation
Availability:	Unrestricted
Abstract:	Optical flow, also known as image flow, is of fundamental importance in the processing of image sequences. Various approaches to determine optical flow have been proposed --- others continue to appear. Using several sets of real image sequences, here, the results of three different approaches to the the determination of optical flow are reported, the approaches are the Gradient-based approach, the Correlation-based approach and the Correlation-feedback approach. The comparisons are primarily empirical, and they show that the performance of these different algorithms differs significantly among the different image sequences, indicating that there does not exists one superior algorithm which could be suitable to every different kind of situations. Every individual algorithm has its own advantages and limitations. Also, recent research shows that motion analysis has many potentials in the computer vision area. There are basically two different approaches to recover structure of objects and relative motion between objects and cameras: one is the optical flow field approach and the other is the feature correspondence approach. The unified optical flow field (UOFF) [5] is a generalization of the optical flow to stereo imagery. Here, through the results obtained from the real image sequences, the feasibility of U01,1- approach to motion analysis has been shown.