Skip to main content
ARS Home » Plains Area » Las Cruces, New Mexico » Cotton Ginning Research » Research » Publications at this Location » Publication #74244
Title: IMAGING SYSTEM DESIGN CONSIDERATIONS FOR THE MACHINE VISION NOVICE

Author
item Lieberman, Michael

Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/29/1996
Publication Date: N/A
Citation: N/A

Interpretive Summary: Machine vision (MV) is now an important tool in the grading of agricultural products. This paper presents situations that frequently cause problems to first time designers of video imaging systems. The issues presented are usually apparent after some part of the MV system does not produce reasonable, expected results. Problems discussed relate to lighting, optics and sensor choice.

Technical Abstract: Machine vision (MV) is now an important tool for grading agricultural products. Although good tests exist describing MV system components, there is usually a gap between textbook descriptions and implementation. Questions received from researchers beginning MV projects and subsequent discussions have resulted in directing this article toward those planning their first system. It is assumed the reader has minimal experience but has access to resources discussing basic MV techniques. The issues presented here are usually apparent after some part of the MV system does not produce expected results. The researcher must discover the flaw in the system's design. Some of the problems presented were found this way. Other situations were recognized as requiring special considerations, but the extent of the problems or solutions were not obvious. The reader's situation will probably vary from the author's, but should this paper aid the reader in recognizing and circumventing a problem, the author's purpos will be met. Since system components impact how real-world optical information is changed or distorted while being converted to digital information, some components are described on a conceptual level. A better understanding of these components can help the researcher more effectively design the MV system. The example machine vision system described was designed to categorize different types of foreign matter in cotton samples.