Scene Matching by Hierarchical Correlation,
Abstract
In this paper the authors present an implementation of hierarchical scene matching in the VISIONS image processing cone - a pyramidal processing architecture. The problem of scene matching is common to many applications in machine vision including registration, motion detection, and stereo vision. Scene matching by feature correlation can solve this problem but suffers from computational expense and failure in highly textured images. Hierarchical correlation provides both a cheaper matching algorithm and a coarse-to-fine matching strategy that overcomes textural problems by matching on gross image structures first. These methods fit naturally into the processing cone or pyramid architectures that have been proposed for image processing. Presented is a discussion of the architecture of the processing cone, the construction of image pyramids, and the use of these pyramids in hierarchical correlation. A set of experiments illustrates the operation of these ideas.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1983
- Accession Number
- ADP001212
Entities
People
- Frank Glazer
- George Reynolds
- P. Anandan
Organizations
- University of Massachusetts Amherst