Biological Systems Engineering, Department of

 

Document Type

Article

Date of this Version

1996

Comments

Published in Applied Engineering in Agriculture VOL. 12(4):487-492. Copyright © 1996 American Society of Agricultural Engineers. Used by permission

Abstract

A corn kernel classification procedure was developed in the frequency domain using a two-dimensional Fourier Transform for inspection of stress cracks. Investigations were also conducted to define suitable conditions and optimum image resolution for viewing stress cracks in corn kernels using a computer vision system. A pre-processing procedure included contrast enhancement, edge enhancement, and kernel edge elimination to improve stress crack recognition. A Fast Fourier Transform algorithm was applied to the pre-processed images, and the transformation results were condensed into 33 feature signatures representing position or orientation invariant morphological features. A multivariate discriminant analysis and multiple regression analysis were used to develop classification criteria for stress crack inspection. Both methods were able to detect stress cracks satisfactorily with an average success ratio above 96%.

Share

COinS