Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
Dissertation_Al-SHAHRI.pdf (51.4 MB)
ETD Abstract Container
Abstract Header
Line Matching in a Wide-Baseline Stereoview
Author Info
Al-Shahri, Mohammed
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=osu1376951775
Abstract Details
Year and Degree
2013, Doctor of Philosophy, Ohio State University, Geodetic Science and Surveying.
Abstract
Matching is a fundamental problem in photogrammetry and computer vision. Matching primitives such as points and lines is the most common feature. While point-feature matching received a lot of attention during the last decade, line matching still lacks well-established algorithms. Two different algorithms are proposed in this dissertation. A bottom-up approach, which starts the solution locally and verifies it globally. The second approach is a top-down approach, where the solution begins globally and is confirmed through local constraints. In this dissertation, we attempt to develop a new reliable line matching algorithm across multiple images. The bottom-up approach uses the configuration of different sets of lines in both views, which are referred to here as a mesh. This establishes putative correspondences by evaluating the geometric errors between two different meshes across pairviews. The top-down proposed algorithm exploits the epipolar geometry and coplanarity constraints. Generally speaking, the epipolar geometry is not directly suitable for the line matching problem due to the inconsistency between the endpoints of corresponding lines. In this proposed algorithm, however, we use it as a global geometry constraint to guide the line matching based on the fact that intersections of coplanar lines are preserved across images. This observation is used to obtain a set of candidate correspondences. The candidate coplanarities are then verified via local homographies derived from neighboring point correspondences. The proposed line-matching methods rely only on geometric relations and do not use appearance during the matching process. While the first approach's performance showed poor matching results, the results derived using the second approach showed high matching performance under different viewpoint changes. The successful comparison with the state-of-the-art methods showed the effectiveness of the proposed method on different datasets.
Committee
Alper Yilmaz, Prof (Advisor)
Dorota Grejner-Brzezinska, Prof (Committee Member)
Ralph von Frese, Prof (Committee Member)
Pages
143 p.
Subject Headings
Civil Engineering
;
Computer Science
;
Engineering
Keywords
line matching
;
wide-baseline
;
epipolar geometry
;
coplanarity
;
image features
;
image registration
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Al-Shahri, M. (2013).
Line Matching in a Wide-Baseline Stereoview
[Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1376951775
APA Style (7th edition)
Al-Shahri, Mohammed.
Line Matching in a Wide-Baseline Stereoview.
2013. Ohio State University, Doctoral dissertation.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=osu1376951775.
MLA Style (8th edition)
Al-Shahri, Mohammed. "Line Matching in a Wide-Baseline Stereoview." Doctoral dissertation, Ohio State University, 2013. http://rave.ohiolink.edu/etdc/view?acc_num=osu1376951775
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
osu1376951775
Download Count:
214
Copyright Info
© 2013, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.