Files in this item



application/pdf9124440.pdf (4MB)Restricted to U of Illinois
(no description provided)PDF


Title:Detection of stress cracks in corn kernels using machine vision
Author(s):Kim, Chul-Soo
Doctoral Committee Chair(s):Reid, John F.
Department / Program:Agricultural and Biological Engineering
Discipline:Agricultural Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Engineering, Agricultural
Engineering, Electronics and Electrical
Abstract:Maintaining high quality of corn is very important to both corn producers and buyers. The detection of stress cracks remains one of the most important tasks in corn quality inspection. Such an index of quality would be helpful in assessing not only the end-use values of the corn but also the drying method used and the appropriateness of subsequent handling procedures.
For automatic detection of corn stress cracks, a machine vision system was developed, which simulates the processes that the human visual system uses to perceive the stress cracks from the corn kernel in the conventional candling method.
The automatic stress crack detection system consisted of four consecutive stages and was configured in various ways by selecting different image processing algorithms in each stage. Several edge detection algorithms suitable for inclusion in the automatic stress crack detection system were developed and analytically evaluated. From analytical evaluation, it was found that edge detection algorithms had different characteristics in their responses and that proper threshold values should be assigned to each algorithm. The proper threshold values were selected by a statistical design procedure.
A set of performance criteria also was developed to evaluate the automatic stress crack detection system and used to compare the performance of different configuration on several varieties of corn samples against human inspectors. Evaluation results showed that the system configured with the circular band operator, the Duda road operator, and the Hough transform, performed best; with success rates of 78.2% and failure rates of 8.2%. The performance measures of the system with this configuration were superior to that of human inspectors. When the system was used to distinguish cracked-kernels from sound kernels, its accuracy was higher than 90%.
Issue Date:1991
Rights Information:Copyright 1991 Kim, Chulsoo
Date Available in IDEALS:2011-05-07
Identifier in Online Catalog:AAI9124440
OCLC Identifier:(UMI)AAI9124440

This item appears in the following Collection(s)

Item Statistics