|Butts, Christopher - Chris|
Submitted to: American Peanut Research and Education Society Abstracts
Publication Type: Proceedings
Publication Acceptance Date: 4/15/2006
Publication Date: N/A
Citation: N/A Interpretive Summary: A machined vision system was calibrated and used to identify unshelled peanuts with three or more kernels. Characteristics such as the number of bumps, the area, length, and perimeter of the unshelled peanut were used in the identification. The system was able to accurately identify 90% of the unshelled peanuts with three or more kernels within the unopened pods. Correctly identifying pods containing 3 or more kernels will provide the ability to sort these pods and concentrate into peanuts marketed at a premium price as a niche market of the roasted in-shell peanut trade.
Technical Abstract: Separation of unshelled peanuts containing 3 or more kernels and then niche marketing them can potentially increase the value of unshelled peanuts and thus the profit of peanut producers or processors. Effective identification of peanut pods with 3 or more kernels is a critical step prior to separation. In this study, a machine vision system was teamed up with neural network technique to discriminate unshelled peanuts into two groups: one with 3 or more kernels and the other with two or less kernels. A set of physical features including the number of bumps, area, length and perimeter were extracted from the images taken. These features were used to train an artificial neural network for discriminating the peanuts. Results obtained so far show that the discrimination accuracy of this system for peanut pods with 3 or more kernels was 90.3 %.