Click on “Download PDF” for the PDF version or on the title for the HTML version.


If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

CLASSIFICATION OF CEREAL GRAINS USING MACHINE VISION: III. TEXTURE MODELS

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  Transactions of the ASAE.  VOL. 43(6): 1681-1687 . (doi: 10.13031/2013.3068) @2000
Authors:   S. Majumdar, D. S. Jayas
Keywords:   Texture, Image processing, Grains, Wheat, Barley, Oats, Rye, Durum, Automation

A digital image analysis (DIA) algorithm was developed to facilitate classification of individual kernels of Canada Western Red Spring (CWRS) wheat, Canada Western Amber Durum (CWAD) wheat, barley, oats, and rye using textural features of individual grains. The textural features of individual kernels were extracted from different colors [i.e., red (R), green (G), or blue (B)] and color band combinations [i.e., black&white {(R + G + B)/3}; (3R + 2G + 1B)/6; (2R + 1G + 3B)/6; or (1R + 3G + 2B)/6] of images to determine the color or the color band combination that gave the highest classification accuracies in cereal grains. Of the 25 textural features used in the discriminant analysis, 10 were gray level co-occurrence matrix (GLCM) features, 12 were gray level run length matrix (GLRM) features, and the remaining 3 were gray level features. To reduce the computational time of the algorithm, the original gray level value (250) was reduced to 32, 16, 8, or 4 gray level values and the textural features extracted from each case were used for classification, and the results were compared. The textural features extracted from the green color band at maximum gray level value 8 gave the highest classification accuracies in cereal grains. Using the 15 most significant features in the texture model, the classification accuracies of CWRS wheat, CWAD wheat, barley, oats, and rye were 85.2, 98.2, 100.0, 100.0, and 76.3%, respectively, when tested on an independent data set (total number of kernels used was 10 500). When the model was tested on the training data set (total number of kernels used was 31 500), the classification accuracies were 87.0, 95.7, 100.0, 100.0, and 81.8%, respectively, for CWRS wheat, CWAD wheat, barley, oats, and rye.

(Download PDF)    (Export to EndNotes)