Purity identification of maize seed based on color characteristics

Abstract

In order to identify miscellaneous seed from maize seed accurately and rapidly, maize seed purity identification method based on color extracted from the images of both the maize crown and the maize side was proposed for improving maize seed purity. Firstly, segmentation and single extraction were carried on the original image; secondly, the color models RGB and HSV were used to extract multidimensional eigenvectors from the maize crown and the maize side; finally, multidimensional eigenvectors were projected into one-dimensional space through applying Fisher discriminant theory and K-means algorithm was carried on the new color space. The experimental results show that K-means algorithm based on one-dimensional space received through Fisher discriminant theory can effectively identify maize seed purity, and the recognition rate was over 93.75%. © 2011 IFIP International Federation for Information Processing.