A Pattern-Theoretic Formulation of Shape in R3
Abstract
The goal of this grant was the formulation of a theory of shape in the plane with the theory being applied to problems in image processing. This development was completed in the case of object recognition/image restoration where the planar objects are characterized by their one-dimensional boundaries, such as in many biological shapes. For example, given a two-dimensional 'noisy' image from some technology, such as a visible-light camera, the methods allow me to build in a priori information concerning the object (or objects) in the image, resulting in an a posteriori probability measure on the possible objects in the scene. This approach is designed to capture the structure of those objects which have a large number of degrees of freedom in shape while all still maintain a certain common form; this invariance is not captured by the traditional geometries.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 31, 1988
- Accession Number
- ADA224872
Entities
People
- Daniel M. Keenan
Organizations
- University of Virginia