<![CDATA[The most important thing of production result of fruit is quality. Especially in citrus, it is related to selling value. 92% production of citrus is “Keprokâ€. But now the quality classification in the fruits industry is still done manually, so it becomes subjective. Information technology is needed to speed up the process of quality classification and make it an objective. This research utilizes the extraction feature of gray level co-occurrence matrix (GLCM) citrus image for quality classification. Begins with collecting data of citrus. There are 100 image data, 60 as training data and 40 as test data. Of each training data, obtained one 64x64 pixels good and bad data image. Do pre-processing on the image and GLCM matrix is formed in direction 0°, 45°, 90° and 135°, feature extraction are contrast, homogeneity, energy and entropy. Support vector machine (SVM) is used for good and bad image identification, to get the percentage of fruit defects. The quality classification is Super Grade, Grade A and Grade B. The result shows that the best classification accuracy is 82.5%, with the amount of training data is 20, distance=2 at 45° GLCM.]]>
Copyrights © 2018
