Visual Objects and Environments: Capture, Extraction, and Representation
Tan, Kar-Han
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/81619
Description
Title
Visual Objects and Environments: Capture, Extraction, and Representation
Author(s)
Tan, Kar-Han
Issue Date
2003
Doctoral Committee Chair(s)
Ahuja, Narendra
Department of Study
Computer Science
Discipline
Computer Science
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Computer Science
Language
eng
Abstract
The availability of affordable computational power and graphics rendering capabilities is enabling the creation of realistic imagery that are widely used today in special effects and animation. New forms of visual media with a higher degree of interactivity, such as video games and virtual reality (VR), have also emerged as a result of these technological advances. While there are many factors that contribute to the presentation effectiveness of a VR simulation or video game, one of the most important is visual realism, and the use of images as textures is often the key. In this dissertation we introduce a number of tools and techniques that are aimed at making it easier to create photorealistic virtual objects and environments. In the first part we describe an algorithm for extracting object boundaries from images, formulated as a probabilistic alpha channel estimation problem. The closed-form solution proposed enables detailed and possibly diffused object boundaries to be found, so that visual objects embedded in digital images can be extracted. We also describe an interactive tool that allows this object extraction operation to be performed with loosely drawn freehand sketches. The second part of the thesis introduces an image-based representation for visual objects called facted appearance models that can be used for object recognition, and pose estimation. The model was used successfully in object tracking in video streams, as well as to estimate eye gaze by treating it as a pose estimation problem. The gaze estimation algorithm achieved 0.36 degree accuracy, which is to existing, more complicated eye gaze estimation techniques. The final part of the thesis describes a panoramic camera based on mirror pyramids that is capable of capturing visual environments from multiple view points simultaneously at video rates.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
