Interpreting images of polyhedral objects in the presence of uncertainty
Shimshoni, Ilan Moshe
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/2142/19070
Description
Title
Interpreting images of polyhedral objects in the presence of uncertainty
Author(s)
Shimshoni, Ilan Moshe
Issue Date
1995
Doctoral Committee Chair(s)
Ponce, Jean
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)
Artificial Intelligence
Computer Science
Language
eng
Abstract
This thesis addresses various problems in computer vision which are related to the interpretation of images of polyhedral objects with a special emphasis on the effects of uncertainty. Its contributions span three areas: we first present an approach to the recovery of 3D shape from a single image using line-drawing analysis and complex reflectance models. The algorithm deals explicitly with uncertainty in vertex position. We then propose an algorithm for computing the finite-resolution aspect graph of polyhedral objects. For each region of the aspect graph a representative finite-resolution aspect is computed. Neighboring regions with identical finite-resolution aspects are merged producing the finite-resolution aspect graph. Finally, we develop a probabilistic approach to object recognition. Match hypotheses are ranked by the probability that they are correct. For each hypothesis the pose of the object is recovered and the region of the pose space compatible with the image uncertainty is computed. Hypotheses which match different features of the same model reinforce each other when the corresponding uncertainty regions in the pose space have a non-empty intersection. Sets of consistent hypotheses are ranked by probability that they are the correct interpretation of the features, producing an ordering of the possible interpretations. The three algorithms have been fully implemented and examples are presented.
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.
