Computer Vision Research and Its Applications to Automated Cartography
Abstract
The SRI Image Understanding program is a broad effort spanning the entire range of machine vision research. Its three major concerns are: (1) to develop an understanding of the physics and mathematics of the vision process, (2) to develop a knowledge-based framework for integrating and reasoning about sensed (imaged) data, and (3) to develop a machine-based environment for effective experimentation, demonstration, and evaluation of our theoretical results, as well as providing a vehicle for technology transfer. This report describes recent progress in all three areas. In particular, we have shown that fractal functions are an effective tool for representing natural shapes and provide a good basis for recovering 3-D shape from the shading and texture in a single image. For scenes containing man-made objects, we have found ways of using straight edges to recover the 3-D orientation of surfaces from a single view, and reason about the shape of an object from partial information in multiple views. We have built a new and powerful LISP-machine-based environment for use in image understanding research, and are putting together high- performance systems for stereo compilation, feature extraction, and linear delineation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1984
- Accession Number
- ADA146668
Entities
People
- M. A. Fischler
Organizations
- SRI International