Download the full-sized PDF
Permanent link (DOI): https://doi.org/10.7939/R3KW37
This file is in the following communities:
|Graduate Studies and Research, Faculty of|
This file is in the following collections:
|Theses and Dissertations|
Shape-Guided Interactive Image Segmentation Open Access
- Other title
interactive image segmentation
- Type of item
- Degree grantor
University of Alberta
- Author or creator
- Supervisor and department
Zhang, Hong (Computing Science)
- Examining committee member and department
Mandal, Mrinal (Electrical and Computer Engineering)
Acton, Scott (School of Engineering and Applied Science, University of Virginia)
Ray, Nilanjan (Computing Science)
Jagersand, Martin (Computing Science)
Department of Computing Science
- Date accepted
- Graduation date
Doctor of Philosophy
- Degree level
This dissertation contributes to developing shape-guided algorithms for interactive image segmentation. Prior knowledge which describes what is expected in an image is the key to success for many challenging applications. This research takes advantage of prior knowledge in terms of shape priors, which is one of the most common object features, and user interaction, which is a part of many segmentation procedures to correct or bootstrap the method.
In this research, shape-guided algorithms are developed for different types of interactive segmentation: initial segmentation, dealing with certain types of under-segmentation and over-segmentation mistakes, and final object boundary refinement. First, the adaptive shape prior method is developed in the graph cut framework to incorporate shape priors adaptively. After obtaining the initial segmentation, to deal with under-segmentation due to object fusion, the clump splitting method is proposed to take the advantage of shape information on the bottleneck position of the clumps. For over-segmentation which requires merging, the interactive merging method is implemented. Subsequently, to refine the incorrectly segmented object boundaries, the shape PCA method is developed to utilize statistical shape information when intensity information is inadequate. Shape information is embedded as the key in each of the proposed algorithms throughout the whole segmentation process.
To integrate these proposed algorithms together, a comprehensive interactive segmentation system is developed which embeds five decisive tools: addition, deletion, splitting, merging and boundary refinement. By combining these tools, a state-of-the-art shape-guided interactive segmentation system can be constructed which is capable of extracting high quality foreground objects from images effectively and efficiently with minimal amount of user input.
- Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.
- Citation for previous publication
Hui Wang, Hong Zhang, and Nilanjan Ray. Clump splitting via bottleneck detection
and shape classification. Pattern Recognition, January 2012.Hui Wang and Hong Zhang. Adaptive shape prior in graph cut segmentation.
In IEEE International Conference on Image Processing, September 2010.Hui Wang and Hong Zhang. Improving image segmentation via shape PCA
reconstruction. In IEEE International Conference on Acoustics, Speech
and Signal Processing, March 2010.
- Date Uploaded
- Date Modified
- Audit Status
- Audits have not yet been run on this file.
File format: pdf (Portable Document Format)
Mime type: application/pdf
File size: 6246894
Last modified: 2015:10:12 19:29:50-06:00
Original checksum: 83e39138f13cc5eebaf79c250d886077
Well formed: true
File title: 1 Introduction
Page count: 136