Deformable Topological Templates for Image Analysis.
Abstract
The main achievements of this proposal has been the introduction of a new family of global but sparse image features consisting of geometric arrangements of local responses. These features can serve as building blocks for template models, allowing for explicit modeling of the variability of the object family, and efficient computation of the template match. These features can also be accessed through decision trees to provide very accurate and efficient shape classifiers.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 10, 1996
- Accession Number
- ADA316810
Entities
People
- Yali Amit
Organizations
- University of Chicago