Files and links (1)
PDFOpen Access (License Unspecified), Open Access
Dissertation
Affine invariants for image registration and object recognition using cross-weighted moments and convex hull
Doctor of Philosophy (Ph.D.), Drexel University
1996
DOI:
https://doi.org/10.17918/00008522
Abstract
A new class of global relative and absolute affine invariants is presented for dealing with the problem of affine invariant image matching and registration from discrete scattered data. These affine invariants are derived based on a novel concept of cross-weighted moments which use an affine invariant cross-weight function with a fractional weight factor. The fractional weight factor makes it possible to arrive at low order (zeroth order) affine invariants which are more robust than the higher order regular moment affine invariants. The affine invariant cross-weight function directly exploits the area invariance of the affine transformation. For object matching, a classifier based on a Euclidean distance error measure between the cross-weighted moment affine invariants of the original image and those of the affine transformed test image is used. From the zero and first order cross-weighted moments, we also obtained unique closed-form solutions to the affine transformation parameters without requiring any feature point correspondence information. The equations used to find the affine transformation parameters are linear algebraic. For the partial scene matching and image registration problem, a novel class of local affine invariants derived from the convex hull of the scattered feature point set is derived. They also make direct use of the area (volume) invariance property associated with the affine transformation, and share the attractive properties of the convex hull, namely, the affine invariance, the local controllability, and the data reduction properties. Because of the local nature of these invariants, the appearance and/or disappearance of parts of the scene are effectively dealt with. All the proposed methods and procedures have been tested on a variety of synthesized and real image data and been found to be robust and encouraging.
Metrics
37 File views/ downloads
18 Record Views
Details
- Title
- Affine invariants for image registration and object recognition using cross-weighted moments and convex hull
- Creators
- Zhengwei Yang
- Contributors
- Fernand S. Cohen (Advisor) - Drexel University, Drexel University (1970-)
- Awarding Institution
- Drexel University
- Degree Awarded
- Doctor of Philosophy (Ph.D.)
- Publisher
- Drexel University; Philadelphia, Pennsylvania
- Number of pages
- xi, 99 pages
- Resource Type
- Dissertation
- Language
- English
- Academic Unit
- College of Engineering (1970-2026); Drexel University
- Other Identifier
- 991021888982704721