Vision Algorithms and Psychophysics
Abstract
Representing shapes in a manner suitable for recognition has been a challenge for machine vision. Here we approach this problem by combining studies of representations used by the human visual system with computational studies of how such representations can be derived and manipulated by machine. Both axial- based and contour-based descriptors were investigated, with emphasis on the role of curvature which was found to be an important primitive underlying both types of representations. Related, but unreported, studies include color and motion, which often serve as the glue that allows one to form appropriate groupings of broken image contours or tokens. This research has yielded over fifty publications, with only the major thrust summarized here. Keywords: Image understanding, Shape recognition, Visual pattern recognition, Visual psychophysics, Vision algorithms.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1989
- Accession Number
- ADA216473
Entities
People
- Whitman Richards
Organizations
- Massachusetts Institute of Technology