Integrated approach to three-dimensional motion analysis and object recognition
Leung, Mun Keung
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/2142/21175
Description
Title
Integrated approach to three-dimensional motion analysis and object recognition
Author(s)
Leung, Mun Keung
Issue Date
1991
Doctoral Committee Chair(s)
Huang, Thomas S.
Department of Study
Electrical and Computer Engineering
Discipline
Electrical Engineering
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Engineering, Electronics and Electrical
Language
eng
Abstract
In this thesis, an integrated system for three-dimensional (3-D) motion analysis and object recognition with noisy outdoor stereo images as inputs is presented. The goals are to obtain the 3-D motion description and the identification of the object on the input stereo images. In order to accomplish the desired goals, the system consists of four stages to extract the required information. These four stages are (i) motion estimation, (ii) distinctive feature extraction, (iii) model database, and (iv) object recognition.
The motion estimation is based on 3-D point correspondences which can be derived from matched points on the stereo images. In this stage, we obtain the following: (1) motion parameters, (2) 3-D centroid locations of the object and (3) region of interest (region of the object projected on an image).
With the noisy outdoor stereo images of a vehicle as inputs, there are two features that can be extracted consistently from them. These two features are the region of interest and the wheel pattern of a vehicle. The first feature, region of interest, can be obtained from motion estimation. The second feature, wheel pattern, is extracted in the stage of distinctive feature extraction which involves template matching and the Hough transform. On the other hand, the two corresponding features of a model library can be obtained from its perspective view generated by the model database.
For object recognition, the two features mentioned above are used to describe an object. The description consists of a set of attribute lists which includes (i) region of interest, (ii) number of wheels, (iii) locations of wheels and (iv) size of wheels. By comparing the attribute list set of the image with that of each model library, a confidence measure can be computed. The vehicle on the input images is identified to be one of the model libraries which has the highest value in confidence measure.
The system is applied to sets of stereo image pairs. From the experimental results, the system can successfully provide the 3-D motion description and the identification of the vehicle on all the stereo image pairs.
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.
