Browse Prior Art Database

Imprecision in Computer vision Disclosure Number: IPCOM000131526D
Original Publication Date: 1982-Aug-01
Included in the Prior Art Database: 2005-Nov-11
Document File: 12 page(s) / 43K

Publishing Venue

Software Patent Institute

Related People

Ramesh Jain: AUTHOR [+4]


Wayne State University

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 9% of the total text.

Page 1 of 12


This record contains textual material that is copyright ©; 1982 by the Institute of Electrical and Electronics Engineers, Inc. All rights reserved. Contact the IEEE Computer Society (714-821-8380) for copies of the complete work that was the source of this textual material and for all use beyond that as a record from the SPI Database.

Imprecision in Computer vision

Ramesh Jain and Susan Haynes,

Wayne State University

Exploiting the uncertainty in data and using a model that parallels the human reasoning process can Improve Images In computer vision systems.

Visual perception has perplexed researchers in philosophy and psychology for centuries, and today it is also perplexing computer scientists. Despite persistent efforts by many noted experimenters, we are still not certain how our brain "sees" the visual signals received by our eyes. Some researchers believed that a good understanding of optics, the retinal image, and the anatomy and physiology of eye and brain would unravel the puzzle of visual perception, but the many advances in these fields have not solved the problem.

The last two decades have seen a growing interest in building computer systems that can describe a scene in terms of the objects in the image and the spatial relations among them. The goal of these systems is to obtain 3- dimensional information about the scene from its twodimensional projection, the image.

Many factors, such as illumination, surface properties of objects, geometrical shape, viewing angles, and occlusion of an object by its own parts or by some other objects, convolve in image formation. Moreover, the prep jection from a 3-D world to a 2-D image is many to one, making the recovery of the 3-D information in terms of objects and relations among objects more difficult. Researchers in computer vision recognize that imprecise information from many sources must be combined to understand the contents of an image. ~~3

Difficulties caused by complexity and uncertainty of data may be solved to some extent by using probability and statistics, but in many situations, imprecision is the issue rather than uncertainty. Statistics are merely a part of the ever growing precise mathematical analysis of complex processes, that has resulted in over-formalization in many fields. This complexity led Zadeh to propose the "fuzzy set" theory.4~5 He states,

In general, complexity and precision bear an inverse relation to one another in the sense that, as the complexity of a problem increases, the possibility of analyzing it in precise terms diminishes. Thus "fuzzy" thinking may not be deplorable, after all, if it makes possible the solution of problems which are much too complex for precise analysis.5

Scientists and philosophers have long considered this approximate reasoning a viable analytic tool.6 In this article we show how fuzzy set theory can be used to solve some problems i...